MasterSu: The Sustainable Development of Su Embroidery Based on Digital Technology

Abstract

Su embroidery, as an intangible cultural heritage of China, is a treasure accumulated by human civilization, but it has been gradually fading from people’s view in recent years. To handle the problems of slow creative output, high learning difficulty, and low production efficiency, and to promote the sustainable development of Su embroidery, this study builds an automatic generation system of Su embroidery called MasterSu, based on the CorelDraw platform. The system can automate the generation of embroidery sketches through area texture filling and color recognition, which allows users to participate in the design process. Finally, the performance and usefulness of the system are verified through user experiments, and it is confirmed that the system can facilitate novice users to understand the embroidery culture, learn the embroidery techniques, and create their embroidery works through the system.

1. Introduction

The concept of sustainable development was introduced by the World Commission on Environment and Development (WCED) [1] in 1987, with ecology, economy, and society identified as the three pillars of sustainable development. It was not until 2015, when the United Nations General Assembly adopted the Transforming Our World: The 2030 Agenda for Sustainable Development, that culture was included for the first time in the 2030 Agenda, as the fourth pillar of sustainable development [2]. Culture is crucial to the realization of sustainable societies [3], but with accelerated urbanization and modernization, it is increasingly difficult for intangible cultural heritage to gain a foothold in today’s society due to its “immaterial” nature, and it even faces the risk of gradually disappearing. In 2003, after the Convention for the SafeGuarding of the Intangible Cultural Heritage was issued, UNESCO put forward the concept of safeguarding intangible cultural heritage, and the protection of ICH began to attract the attention of various countries [4]. Furthermore, as a common cultural asset in the development of human civilization, intangible cultural heritage (ICH) has important historical significance, and the continuation of ICH is crucial to the sustainable development of culture.

The importance of safeguarding intangible cultural heritage for sustainable development has been widely recognized by the international community [5]. Chinese Su embroidery, as one intangible cultural heritage, is essential to maintain cultural diversity and sustainability [6]. Its history can be traced back over 2500 years [7]. Su embroidery is a Jiangsu embroidery with production centers around Suzhou and Wu County, and it was included in the first version of China’s intangible cultural heritage protection list in 2006 [8]. Su embroidery is ranked first among the the four famous embroideries because of its beautiful patterns, elegant colors, rich stitching, and fine embroidery work [9,10]. As an intangible cultural heritage, Su embroidery is a cultural treasure that was accumulated during the development of human civilization and has received more and more attention in the sustainable development of modern society [11]. Though cultural heritage has value and importance in human life, it is always in danger of disappearing, and actions and measures are urgently needed to promote the sustainable development of Su embroidery [2].

With the development of the times, the inheritance of traditional Su embroidery has encountered many difficulties, and the industry has been gradually declining in recent years [12]. The reasons for this are as follows: (1) In terms of design creativity, the effect of embroidery is greatly influenced by individual techniques, and the embroiderer’s literacy and emotional expression can affect the artistic effect of the work [13], resulting in a slower process of creative design output. (2) In terms of technique inheritance, it is difficult for ordinary people to participate in the process of making Su embroidery. (3) In the production process, Su embroidery relies on manual work, while the long production cycle and slow efficiency of traditional handmade production cannot meet the procurement needs of the contemporary market [14]. Therefore, the sustainable development of Su embroidery has problems such as slow creative output, high learning difficulty, and low production efficiency.

With continuous development in informatization and globalization, digital technology can transform information into measurable numbers to better transmit and disseminate intangible cultural heritage in a way that is easy to inherit and spread [13]. Sophisticated digitization aids have been developed for intangible cultural heritage, such as a digitization tool for fabric pattern design by Haiying Zhao et al. [15]. Kuo et al. [16] proposed a machine-embroidery-image color analysis system that could automatically and accurately identify and crop out images of repeated patterns in machine embroidery to improve the efficiency of pattern recognition during embroidery. Enabling new productivity in the information era, digitization has become a fundamental condition for the inheritance and development of traditional culture in the information era [17], and digital technology can provide a broader user market for embroidery, thus promoting the inheritance and dissemination of Su embroidery and the sustainable development of intangible cultural heritage. However, the actual stitching steps are different from the rough sketches on the fabric, and stitching in Su embroidery often requires combining several pattern types of stitches, which is a challenging task for beginners [18], thus affecting the inheritance and development of Su embroidery.

This paper develops MasterSu, a Su embroidery digitizing software to help general users learn to embroider. By using our software, users are free to choose and draw a pattern, preview the stitch effects of Su embroidery, and easily and quickly view the final design. By reducing the difficulty of generating ideas for Su embroidery and the difficulty of learning for ordinary users, the developed software promotes the inheritance of Su embroidery in society and culture and its dissemination in the market, as well as the sustainable development of intangible cultural heritage.

2. Related Works

2.1. Research on the Sustainability of Intangible Cultural Heritage: The Example of Su Embroidery

More and more scholars have begun to conduct research with the goal of the sustainable development of ICH, and they are devoted to exploring more effective methods for the sustainable development of ICH. Among these methods, digital preservation strategies for ICH have received much attention [19,20]. The Law of the People’s Republic of China on Intangible Cultural Heritage promulgated in 2011 defines the “protection” of ICH as the “inheritance and dissemination” of representative items. It indicates that inheritance and dissemination have become the two main focuses of China’s intangible cultural heritage protection. However, the inheritance and dissemination of Suzhou embroidery are difficult to protect well. Inheritance is the fundamental way for the continuity and development of traditional culture [17], and cultural inheritance is also a process of cultural reproduction [21]. The “tangibility” of Su embroidery makes it a carrier of history and culture and a witness and epitome of cultural inheritance. The patterns and matching colors of Su embroidery have distinct national characteristics and rich local colors. However, the process of the inheritance of embroidery is impeded by slow creative output and the high learning difficulty, which leads to the shrinkage of the embroidery population and makes the inheritance and promotion of embroidery difficult. The circulation of embroidery in the market is equivalent to the spread of national and local culture in society, which is constrained by high learning difficulty. In addition, the propagation of Suzhou embroidery, i.e., the movement and change of things in time and space, is challenging. However, the low production efficiency of Su embroidery leads to the slow circulation of Su embroidery in the market, so the problem of low production efficiency needs to be solved to promote the spread of Su embroidery. Traditional culture is the precipitation of civilization in the historical dimension of a country, nation, or region, and it is the source of spiritual strength and an important basis of identity for a nation [22].

The development of Su embroidery’s heritage in contemporary society has been required to be both culturally transmitted and commercially profitable. Its development is not only influenced by generations of inheritors, but also driven by the preferences of consumers, and the constraints of production developers. The users of the computer-aided Su embroidery design platform can be divided into three layers: the internal layer, the production layer and the external layer. Of these, the internal layer refers to the inheritors of Su embroidery, whom can be divided into ordinary beginners and skilled embroiderers. The production layer refers to manufacturers, wholesalers, retailers, etc. The external layer refers to the people who actually buy and use embroidery, i.e., the consumers of embroidery.

The needs of different stakeholders for the computer-aided Su-embroidery design platform are somewhat different. For the inheritors, the computer-aided Su-embroidery design platform enables more people, in general, to participate in learning Su embroidery, so that the culture of Su embroidery can continue to be passed on. For stakeholders such as production manufacturers, wholesalers and retailers, the computer-aided Su-embroidery design platform can give quick sketches of embroidery and improve production efficiency. At the same time, Su embroidery, as a symbol of a local culture, helps to strengthen the ties between local people and communities, and can also bring a thriving economy to the local industry. For the consumer, the computer-aided embroidery design platform can help the user to see how the embroidery will look in advance by generating previews and choosing a style that better suits their preferences, reducing waste of resources and driving economic consumption at the same time. Overall, the use of computer-aided design methods to promote the revival and development of embroidery can contribute to cultural, social, economic and environmental sustainability, and help to achieve sustainable development goals [23]. Therefore, this paper will integrate the perspectives of different stakeholders’ needs to build a computer-aided design platform to promote the development of Su embroidery.

2.2. Digital Preservation of Intangible Cultural Heritage

Digital technology is the most urgent and effective way to preserve and transmit intangible culture [24]. There have been many digital-conservation solutions for ICH. The digital-conservation approach for ICH has gained much attention in academia, and a large number of digital-conservation solutions have been proposed for ICH, in three specific directions. The first direction is the digital acquisition of ICH. Physical contents such as documents, materials, and films related to ICH items are digitally recorded and stored [25]. The second direction is about the digital preservation of ICH. Digital preservation of NRMs is carried out by building a digital database or management platform. For example, Wang and Hong proposed to build an information-sharing platform for ICH protection to expand the open platform of ICH information resources and realize information sharing [26]. The third direction is the digital development of ICH.

For movable cultural heritage, computer-aided technology allows the integration of design and technology to create cultural products with cultural heritage and contemporary needs. For example, Cai [27] proposed a tool to complete paper-cutting design and production through a computer-aided system. Guo [24] proposed a model of ICH image recognition to inherit and protect ethnic patterns. For immovable cultural heritage, digital technology is also a proven means of intervention. For example, the detection of the degradation of archaeological sites is a very time-consuming task that requires the assistance of specialist knowledge. Computer vision methods, however, are an effective solution, eliminating both human error and difficulties in the field. Scholars such as Hatir have used the Mask R-CNN algorithm to detect and map the degradation observed in archaeological sites [28,29].

Studies on the digital conservation of both movable and immovable cultural heritage have shown the effectiveness of digital technologies for the conservation of ICH and have provided a wealth of recommendations. However, a key issue for the contemporary ICH is how to integrate a long-standing ICH into today’s society [30]. Digital development of ICH is an effective way to help ICH achieve innovation by integrating modern information technology into ICH culture and transforming cultural connotations into cultural competitiveness to keep up with the times, adapt to the development of modern society, and better transmit and disseminate ICH. The use of digital means can maximize the preservation of embroidery art, and it is highly necessary and significant to design and process it into the aesthetic form of digital media to maximize the display and dissemination of embroidery art [31].

2.3. The Computer-Aided Technology for Embroidery

After more than 2000 years of development, Su embroidery has tended to be programmed and standardized at all levels in pattern, color matching, and stitching, and these regular features have allowed computer-aided tools to intervene [32].

As for the patterns of Su embroidery, there are various representations of figures, pets, flowers, birds, landscapes, etc. The specific picture representation depends on the ideas constructed by the embroiderer during the embroidery process, and the picture representation is unstable. Since people instinctively resist complex information, the process of embroidering a given image on fabric can be a challenging task for beginners, but the intervention of digital technology can make the information simple [33]. Liu et al. [34] proposed a possible solution to develop an intelligent embroidery programming environment that can input an initial drawing into the computer. Then, the outline image is processed into the embroidery outline, thus obtaining a high-quality embroidery sample, which confirms the feasibility of the digital conversion of embroidery patterns. However, Su embroidery’s beautiful and cleverly conceived patterns require a more refined edge extraction of the pattern’s regional layout for subsequent combination and re-creation by the user, when the computer-aided design is performed.

In terms of color matching, Su embroidery is famous for its rich color layers. Su embroidery artists usually use many different color threads of the same color or neighboring colors to match and embroider a natural color effect, and a fine Su embroidery work often uses many color threads of the same color scheme to create a natural transition effect. There are many sophisticated algorithms for color matching design. For example, Wu et al. [35] proposed a probabilistic method to identify the knowledge of qualitative color selection from color schemes and assist in color matching. Hsiao and Yang [36] developed a product color design system for color matching design. However, there is no algorithm research for color matching in embroidery, and an intelligent color-matching system for Su embroidery needs to be developed.

As for stitching, different stitching methods have their own organizational rules and unique expressive effects. With the richness of the subject matter of Su embroidery works, Su embroidery artists have derived new stitches according to changes in the subject matter [37]. Su embroidery makers use different stitches according to different objects to express the texture of the embroidered objects and enhance the artistic expression of Su embroidery. The most commonly used stitch in Su embroidery is the loose set stitch, with its stitches arranged at different heights, flexible organizations, and natural turns. However, the existing programs have been developed for regular stitches. For instance, Jie et al. [38] and Yang et al. [39] explored and developed random stitch embroidery, but no algorithm has been developed for loose-variegated stitch yet.

Therefore, the platform for a computer-aided Su embroidery design needs to focus on three main parts: (1) Pattern: partitioning the given pattern, then extracting the edges of the area with refinement to automatically generate the base pattern for subsequent combination and re-creation by users. (2) Color matching: extracting the color blocks of Su embroidery for parametric processing to generate the base color matching containing gradient colors. (3) Stitch: providing loose-variegated stitch algorithms to reduce the learning difficulty of beginners. This study proposes MasterSu, a computer-aided tool for the automatic generation of Su embroidery patterns, color schemes, and stitches, which can assist users in Su embroidery.

In order to verify that the computer-aided Su embroidery design platform achieves its goal of sustainability, we will consider the following aspects: educational effects, technical effects and inducing effects [5]. Of these, the educational effects refer to whether the technology plays a role in educating the public. The technical effects refer to whether the new technology is appropriately applied to the field. The inducing effects refer to whether the new technology encourages the public to take the initiative in learning about embroidery in order to promote its dissemination.

3. Method

In response to the current problems of inheritance and dissemination of Su embroidery, this study develops MasterSu, a computer-aided Su-embroidery design system, to help ordinary users quickly learn Su embroidery. The system includes three processes: creative input, parameter presetting, and design output. The creative input aims at the user’s independent selection of target objects in the image, ticking the main form, and drawing the embroidery thread direction. Based on the input pattern form, the color information and embroidery parameters in the area are extracted, and the number of embroidery colors, embroidery thread width, embroidery thread length, and other parameters are preset for the embroidery design. The design is digitized according to the embroidery stitches selected by the user, and, finally, the design scheme is output. The main contents of this study are image information extraction, embroidery stitch design, and color information mapping.

3.1. Image Information Extraction

Image is a similar and vivid description or portrait of an objective object, and it is the most common information carrier in human social activities, containing a variety of data information such as the form, color, and structure of the object. The creation based on images is an important way to develop Suzhou embroidery design. Image-based embroidery design mainly obtains the shape parameters and color parameters of the target object in the image. Among the parameters are: (1) Morphological parameters, which include the outline of the object and the texture direction, where the outline constrains the style and scope of the embroidery pattern, and the texture direction determines the trend of the embroidery thread. (2) The color parameter, which gives the embroidery thread color to fit the real state of the object.

3.1.1. Morphological Parameter Extraction

Morphological parameters include object contours and texture orientation. The image contains single or multiple objects, and the user extracts the target object from the image closely related to its aesthetic criteria, perceptual intention, and design requirements. Faced with the complex demands of multiple factors in the user’s decision-making process, the intelligent selection of the target object does not fully meet the user’s selection intention. Therefore, this study adopts a user-selected route to extract the object contour parameters, and the study unfolds based on the 2D flat software CorelDRAW. In this way, the user can quickly complete the contour outline of the target object in the image using the curve tool. The morphological composition of the target object is not a single form but may consist of a combination of several forms. Thus, the user needs to draw all the forms of the object because the outline of the object is composed of multiple curves. The texture of the object is an important medium to show the trend of the texture, such as the texture of the leaf with the leaf vein. In embroidery, the texture of the object is shown by the direction of the embroidery thread, and the user draws the texture curve as the input parameter of the system by combining the object form, color change, and texture change.

3.1.2. Color Parameter Extraction

Color is the main feature of an image, and there are many mature techniques for its extraction, such as the color histogram reflecting the distribution of image color composition [40]. In this paper, the K-Means clustering algorithm is used to extract the characteristic color of the target image. The clustering method of color expresses the color values as three-dimensional (e.g., RGB color) or four-dimensional (e.g., CMYK color) spatial coordinate location points, and the central color coordinates of various color point sets are obtained by several iterations according to the set number of target color categories, i.e., the target color information. The color extraction for embroidery design does not need to traverse all the image colors but only needs to cluster the colors within the outline of the object. In this paper, color extraction consists of three processes: color collection, color screening, and color clustering.

Color-collection stage: All pixels of the image are processed into a set of coordinate points in the color space. Each pixel is coded and its central coordinate value (x_i, y_i) and the corresponding color value (R_i, G_i, B_i) are recorded. The number of color types n is counted according to the color values to determine whether to perform image color clustering. If the number of color types is smaller than the target number, no clustering is required.

Color-screening stage: The purpose of color extraction is to provide a color reference for the target object only, according to the target object outline range. The color-screening stage is divided into two processes: coarse screening and fine screening.

Step 1: Coarse screening extracts the color within the spatial range of the target object, as shown in Figure 1. The method obtains the spatial parameters of the target object, i.e., the spatial range (x_min, x_max, y_min, y_max), after traversing all the colors to obtain the color set for coarse screening.

Step 2: Fine screening collects the color that is completely within the range of the target object, as shown in Figure 2. First, the composition curve of the target object is read, and the curve is segmented according to the color-screening accuracy value A to obtain the range segmentation line set S. The specific method is: obtain the number of sub-paths of the contour curve (i.e., the number of turning points of the contour curve), calculate the length of each sub-path curve L_path-i, calculate the number of sampling points n_path-i based on the length of the path curve and the sampling accuracy, record the coordinates of the sampling points to obtain the sub-path sampling points p_path-i-j, record the first and last points of the segments s(p_path-i-j, p_path-i-j+1) according to the number of segments, collect all the sub-path segments to obtain the set of contour curve segment lines S. Then, all the color points for coarse screening are traversed, the color point O₁(x_i, y_i) is taken as the starting point, and ray O₁P₁ is made over the point P₁(x_i, y_i+l), $l>\sqrt{{\left(x_{max}-x_{min}\right)}^2+{\left(y_{max}-y_{min}\right)}^2}$. In this process, the starting point O₂(x_s-j, y_s-j) and the ending point P₂(x_s-j+1, y_s-j+1) of the segmented line are extracted, ray O₁P₁ and O₂P₂ are combined, and t₁ and t₂ are calculated by the vector fork multiplication in Equation (1). If the formula holds, then the two lines intersect; otherwise, they do not intersect.

$$O_2-O_1=D_1{t}_1-D_2{t}_2$$

(1)

where, D₁ is the vector of O₁P₁; D₂ is the vector of O₂P₂; t₁ and t₂ are the parameters of the intersection points located on O₁P₁ and O₂P₂, and they fall within [0, 1].

The points located within the target range are recorded to form a filtered point set, and the color values of the points are extracted and recoded. The number of colors to be extracted needs to be specified by the user before color clustering, and then the coordinates (i.e., color components) of the point set are clustered.

Color clustering stage. The K-Means color-clustering method has been studied in depth by Liu et al. [41,42], research members of our team, and a brief overview is given here. Based on the filtered point set, the distance between the color value of each pixel in the point set and each cluster center is calculated in turn, and it is assigned to the category represented by the nearest cluster center. The update of the cluster centers is performed iteratively, and the termination of the iteration is determined by the condition that the maximum distance between the calculated cluster center and the last cluster center is less than a fixed threshold, and the judgment formula is shown in Equation (2). The final clustering center is output at the end of the algorithm, which is the extracted color.

$$f_{max}=\sqrt{{\left({C_r}^{\prime }-C_r\right)}^2+{\left({C_g}^{\prime }-C_g\right)}^2+{\left({C_b}^{\prime }-C_b\right)}^2}/\sqrt{3\times {255}^2}, f_{max} <f$$

(2)

where f_max is the maximum distance between the current clustering center and the last clustering center; C_r′, C_g′, and C_b′ are the RGB values of the current clustering center; and C_r, C_g, and C_b are the RGB values of the last clustering center. Furthermore, $\sqrt{3\times {255}^2}$ is the diagonal length of the three-dimensional color space, i.e., the maximum distance. Dividing the maximum distance by the absolute distance is to convert the distance into a relative value located between 0 and 1 for evaluation; f is the judgment threshold, and it is selected based on the minimum difference recognizable to the human eye, which is an empirical value. In this paper, f takes the value of 0.05.

Then, color clustering is performed in RGB mode, and the initial clustering centers are uniformly distributed along the diagonal of the RGB three-dimensional space, i.e., a series of grayscale values from pure black to pure white. The final clustering result obtained is an extracted color vector and a weight vector. The extracted color vector maintains the color value of each extracted color, and the weight vector maintains the percentage of the number of pixels in each color class. The extracted colors corresponding to each form of the target object are obtained after multiple clustering, but the color values and the number of colors in each form region are not the same. For example, the results obtained after clustering similar image parts in Figure 3 are not the same. If all the similar colors are applied to Su embroidery, it will cause color confusion, and the result will not be in accordance with visual aesthetics. For this reason, the secondary clustering of similarly clustered colors is performed as follows. First, the minimum distance of all extracted colors in the three-dimensional color space is calculated and converted into relative distance values. Then, the colors with a threshold of less than 0.05 are screened for clustering, and the color values greater than the set threshold are output. According to the secondary clustering results, the extracted color corresponding to each morph is re-corrected.

3.2. Su-Embroidery Stitch Design

Variegated stitch is one of the traditional stitches of Suzhou embroidery. According to the effect of the pattern, it can be divided into flat-variegated stitch, set-variegated stitch, and loose-variegated stitch. Among them, the loose-variegated stitch is one of the most widely used stitches in Suzhou embroidery. Its main feature is that the outer edge of the first batch of stitches is neat, the stitches are dense, and the inner edge is different in length. Therefore, in this paper, the digitization method of variegated stitches is investigated based on the variegated stitch using the secondary development platform of CorelDraw.

Based on the difference in embroidery methods, there are single-variegated stitch, double-variegated stitch, wood-comb-variegated stitch, set-variegated stitch, flat hair-variegated stitch, and live hair-variegated stitch. From the variegated-stitch embroidery method shown in

Table 1. The detailed explanation and diagram of set stitch embroidery methods.

No.	Variegated Stitch Method	Embroidery Method	Illustrations
1	Single-variegated stitch	The first batch starts from the edge of the needle, the edge of the mouth, neatly; the second batch in the first batch of needle drop, the first batch needs to leave a line of clearance to allow the second batch of needles; the third batch needs to turn into the end of the first batch by a centimeter, and then leave the fourth batch of needle clearance.
2	Double-variegated stitch	The embroidery method is similar to that of a single-variegated stitch but deeper and shorter than a single-variegated stitch. The difference is that the fourth batch is connected to the first batch, i.e., the second batch is connected to three-quarters of the first batch, the third batch is connected to two-quarters of the first batch, and the fourth batch is connected to one-quarter of the first batch.
3	Wooden comb-variegated stitch	The embroidery method is similar to that of a single-variegated stitch but looser and thinner than a single-variegated stitch. The difference lies in the fact that after the first batch comes out of the edge, the second batch drops the stitches at half of the first batch, and every other thread sets a stitch, and the second batch accesses a centimeter of the end of the first batch, and clips in the gap left by the second batch.
4	Set-variegated stitch	The embroidery method is similar to that of a single-variegated stitch, but the difference lies in the fact that the first batch of stitches at the outer mouth is larger, the embroidery batch is less than the number of stitches, and the batch hides short stitches. At the end of the embroidery, the surrounding lines are all concentrated in one eye, and one batch is overlapped with another batch, just like a single-variegated stitch.
5	Flat hair-variegated stitch	The embroidery is performed according to the pattern of the piece of hair, and the hod is the same as the double-variegated stitch, except that the lines are slightly longer or shorter.
6	Live hair-variegated stitch	There are two types of variegated stitch: one is to embroider three stitches flat first, then add a cross-stitch on the third stitch. The second is tight inside and with a radial shape outside, embroidered from outside to inside, following the shape of the animal to turn the momentum, with the thread overlaying on the second variegated stitch falling at half of the previous batch.

, single-variegated stitch, double-variegated stitch, wood-comb-variegated stitch, and set-variegated stitch embroidery methods are similar, and they are the most common embroidery methods. The difference is concentrated in the direction of the embroidery thread arrangement. The flat hair-variegated stitch is a set of embroidered pieces of hair. It is based on the pattern of the piece of hair and is the same as the double-variegated stitch, except that the lines are slightly longer and shorter. The live-hair-variegated stitch is a set of embroidered animals that combines the two embroidery methods according to the shape of the animal. By combining the embroidery method and the embroidery object, this paper classifies the stitch methods into three categories: single-variegated stitch, flat hair-variegated stitch, and live hair-variegated stitch. Among them, the single-variegated stitch is the most commonly used embroidery method, and the latter two are mainly used in smaller areas and with more strict requirements on the precision and drawing method of the target outline, e.g., feathers and hairs. Therefore, in this paper, only the digitized embroidery method of single-set embroidery is studied, and the digitized embroidery stitch method of single-set embroidery is designed by taking the object outline, texture direction, and thread length and width as the basic parameters.

The single-variegated stitch stitching method adopts an outside-in method, alternating short and long stitches. The short and long stitches intermingle to present a gradual embroidery pattern, and the stitches neatly and closely follow the edge of the pattern. The single-variegated stitch and the set-variegated stitch are embroidered with double needles (one long and one short) as a unit; the double-variegated stitch is embroidered with four needles (changing from long to short) as a unit; the wooden-comb-variegated stitch is embroidered with multiple needles (changing from long to short) as a unit with interlocking stitches. The basis of the single-variegated-stitch needlepoint embroidery method is to alternate and overlap each other with long and short embroidery threads as units, presenting excessive stitches. In addition to the basic parameters, the single-variegated stitch method also requires the input of the number of stitches in the unit to classify the overlapping levels and to calculate the starting and ending points of the stitches. Su embroidery is divided into three embroidery processes: the starting layer, the transition layer, and the finishing layer.

The layers of embroidery are divided by the input parameters of the texture curve, the contour curve and the length of the embroidery thread. The range of the starting layer is determined based on the outline of the object and the direction of the texture by first calculating the intersection points p_start and p_end of the outline curve and the texture curve, extracting the length of the intersecting section of the curve L_intersect, and then calculating the relative point pl on the intersecting section to the starting point p_start of the texture curve based on the ratio of the length of the embroidery thread L_standard to the length of L_intersect. Then, the normal angle α of the pl point on the intersecting curve is calculated, a ray is drawn, the contour curve at two points. q₁ and q₂, are intersected, the range contour line s₁ and s₂ and the articulation line s₃ with q₁ and q₂ as division points are extracted. Finally, s₁, s₂, and s₃ form the starting layer, as shown in the yellow area of Figure 4.

3.2.1. Starting Layer

The starting layer is the beginning of the embroidery, and it is used to determine the starting coordinates of the needle and the texture of the embroidery thread. Based on the starting-layer boundary curves (s₁ and s₂) and the preset embroidery-thread parameters (length l and width w), the starting and ending points of the embroidery thread on the boundary curve are calculated in three steps.

Step 1: Using the embroidery line width d as the sampling accuracy, calculate the coordinates of the sampling points on the contour line, integrate m sampling points to obtain the point set P, and divide the number of unit stitches n by the point set to obtain k units. If m is a multiple of k, then they are all complete units; otherwise, the last unit is a non-complete unit.

Step 2: Connect the two points of p_start and p_end and calculate the horizontal angle β of the first section of the texture curve.

Step 3: Calculate the length of the embroidery thread from the longest to the shortest based on the number of unit stitches n. ${\mathrm{l}}_{\mathrm{j}}=\mathrm{l}\ast \left(\mathrm{n}-\mathrm{j}+1\right)/\mathrm{n}$, j = (1, n). Using the point set P as the starting point combined with the angle β, calculate the endpoint of each unit embroidery thread, ${{\mathrm{P}}^{\prime }}_{\mathrm{i}-\mathrm{j}}\left({{\mathrm{x}}^{\prime }}_{\mathrm{i}-\mathrm{j}},{{\mathrm{y}}^{\prime }}_{\mathrm{i}-\mathrm{j}}\right)=\left({\mathrm{x}}_{\mathrm{i}-\mathrm{j}}+{\mathrm{l}}_{\mathrm{j}}\ast \cos \mathsf{\beta },{ \mathrm{y}}_{\mathrm{i}-\mathrm{j}}+{\mathrm{l}}_{\mathrm{j}}\ast \sin \mathsf{\beta }\right)$, where i = (1, k) and j = (1, n). Connect the endpoints to obtain the embroidery unit, as shown in Figure 5.

3.2.2. Transition Layer

The embroidery method of double-variegated stitch differs from single-variegated stitch, set-variegated stitch, and wooden comb-variegated stitch in the number of units of embroidery threads. The former consists of three articulated units, including long and short units, equal-length units, and short length units, as shown in Figure 6a; the latter consists of two articulated units, including long and short length units, as shown in Figure 6b. Therefore, two patterns exist at the articulation of the starting layer and the transition layer. One is equal-length embroidery transition, which articulates equal-length embroidery immediately after the endpoint of the upper layer (from long to short), and then articulates reverse embroidery (from short to long). The other is reverse embroidery transition, which articulates reverse rust directly after the endpoint of the upper layer.

With this regular articulation of the recorded endpoints, there are three types of excesses in the process.

• Stable transition means that the number of upper-layer endpoints is the same as the number of lower-layer start points, as shown in Figure 7a.
• Convergent transition means that the number of endpoints of the upper layer is more than the number of starting points of the lower layer, as shown in Figure 7b, resulting in some embroidery thread lengths smaller than the standard parameters. If the threads intersect, the coordinates of the intersection point are calculated and the starting point is connected to the intersection point; otherwise, the endpoint is calculated with standard parameters and recorded as the starting point of the next layer.
• An expansive transition means that the number of endpoints of the upper layer is less than the number of starting points of the lower layer, as shown in Figure 7c. In this case, the new starting point of the embroidery thread needs to be obtained from the object outline. Then, the endpoint is calculated based on the embroidery thread parameters and the object outline, and it is recorded as the starting point of the next layer.

3.2.3. Finishing Layer

The embroidery thread composition of the finishing layer is different from that of the starting layer and the transition layer in that the embroidery thread parameters are floating. The essential reason is that the endpoint of the embroidery thread in the finishing phase needs to be located on the target outline. The standard parameters cannot be used to constrain the endpoint of the stitching layer due to the irregular shape of the pattern. If the distance between the starting point and the endpoint Di is greater than the standard embroidery thread length D_standard-j, directly connecting the endpoint will change the embroidery line texture, resulting in a poor embroidery effect. Therefore, the starting point needs to be pre-processed as follows.

Step 1: Extract the finishing layer boundary lines s₄, s₅, and s₆, as shown in Figure 4, according to the layered lines and contour lines.

Step 2: Based on the angle ω between the texture curve and horizontal line of the finishing layer, the starting point set Q_start(x_i, y_i) of the finishing layer and the corresponding length parameter set R of the embroidery line, the coordinates of all corresponding endpoints are calculated to obtain the endpoint set Q_end(x′_i, y′_i), and whether the embroidery line intersects with the boundary line of the finishing layer is judged based on the starting and ending points by Equation (1). If they do, then the endpoint is in the floating range, and the intersection point coordinates are recorded; otherwise, the starting and ending points are connected, the starting point of the embroidery line in the finishing layer is updated, the new endpoint is included in the point set Q_start as the starting point, and the coordinates of the corresponding endpoint are calculated based on the new starting point. Then, whether the new start and end points intersect with the boundary is judged again. If they do, the intersection points are recorded; otherwise, the above operations are repeated until they intersect, as shown in Figure 8.

Step 3: Update the starting point set, integrate the intersection points as the endpoint set, and keep the coding correspondence between the two-point sets.

Traverse all the starting points and connect the endpoints to complete the finishing layer of embroidery.

3.3. Color-Information Mapping

The number of pixel frames in the contour range of the object is u. The colors of all pixel frames are traversed, and the corresponding clustered color number of each pixel frame is recorded after clustering. However, due to the complexity of the image color composition, the mapped colors may have a large number of dots, i.e., a small number of different colors in a large area of the same color, as shown in Figure 9a. If the color of embroidery thread is assigned based on the clustered colors with dots, the color of embroidery thread may change abruptly and affect its aesthetics. For this reason, after color clustering, the mapped colors of all pixel frames are reviewed to correct the heterochromatic colors; the process is divided into three steps.

Step 1: Iterate through all the pixel cells and extract the cluster color number corresponding to all the pixel cells within the radius of 2.5 pixels. Then, take the center of the pixel cell as the circle in turn and make a judgment.

Step 2: Extract the cluster color number corresponding to the non-center pixel grid. If the number is the same, it is recorded as “V = true”; otherwise, it is recorded as “V = false”.

Step 3: If “V = true”, then determine whether the cluster color number of the center pixel is the same as the cluster color number of the non-center pixel. If it is the same, then the corresponding cluster color number remains unchanged; otherwise, the cluster color number of the center pixel is corrected to the cluster color number of the non-center pixel. “V = false” is used to judge all the clustering color numbers: if only two colors exist and the minority color position is not longitudinally or horizontally distributed, as shown in Figure 9b, and the percentage is less than or equal to 1/10, then the clustering color number corresponding to the central pixel grid is corrected to the clustering color number of the majority color. If the minority colors are distributed vertically or horizontally, as shown in Figure 9c, the cluster color number of the central pixel cell remains unchanged. If there are more than two types of clustering colors, the clustering color number of the center pixel lattice remains unchanged.

After the color review, the center point T_i(x_i, y_i) of each embroidery thread is extracted, and the distance between the center point and the center of each pixel grid is calculated as (r₁, r₂, …… , r_u). By comparing with r, the nearest pixel grid is found according to the center point, and the corresponding cluster color number is recorded. Then, the cluster color is extracted, and the corresponding color of the embroidery thread is finally determined.

4. Results

According to the above research method and technical route, the development of the digital system of Su embroidery variegated stitch stitches was completed based on the CorelDraw secondary development platform, and the system interface is shown in Figure 10. The system provides four types of embroidery stitches: single-variegated stitch, double-variegated stitch, set-variegated stitch, and wooden-comb-variegated stitch. The functions are mainly composed of morphological-parameters input, source-image color extraction, and auxiliary digital embroidery. This section uses the source diagram of “Persimmon” in Figure 3 to introduce platform functions through examples.

Users can outline the target object form in CorelDraw according to the image, and the outline should be divided into categories according to the object type and texture direction. The morphological outline and texture curve of the target source image are shown in Figure 11, which are divided into 13 areas, and the morphological outline and texture curve are selected in turn to input into the system. The sampling accuracy is set to 100, and the number of extracted colors is set to 3. Then, the source map is selected to input into the system, and the color results are output after color extraction and clustering, as shown in Figure 11. Finally, the color clustering results are displayed at the bottom of the system.

The user then processes the morphological regions in turn, selects the morphological contours and texture curves corresponding to the region, and sets the embroidery stitch and embroidery thread parameters to be used in the region. In order to show the optimal embroidery effect, after several design iterations, we set up areas 1 and 2 with a single-variegated stitch, area 3 with a combination of single and set-variegated stitch, and area 4 with a double-variegated stitch for embroidery. The final output of MasterSu is shown in Figure 12.

5. Evaluation of MasterSu

After development, the system needs to be evaluated before it can be put into use [43], so this paper investigates users’ feelings and evaluation of MasterSu through usability testing. The two main purposes of using usability testing in this study are: (1) to verify whether the system performs well in real usage scenarios and whether the system operation is inconvenient and friendly to beginners; and (2) to verify whether the system can effectively promote the sustainable development of Su embroidery.

5.1. Procedure

Before the experiment started, 10 novice users were gathered, including five male and five female, with an average age of 25 years. The subjects had no experience in Su embroidery and did not know much about Su embroidery before the experiment. Each subject was asked to complete the tasks in

Table 2. Usability testing tasks.

No.	Mission	Paths
Step 1	Viewing the explanation and creation instructions for Su embroidery	Open the software; view Su embroidery explanation; view creation instructions
Step 2	Selecting images for editing	Search image library to select images/custom import images/image search function
Step 3	Contouring	Select the brush; outline the image; confirm
Step 4	Drawing of texture direction	Select the brush—according to the different blocks of the image for the direction of the texture of the drawing
Step 5	Setting material, stitch, and color parameters	Select material; matching array; select length to short line ratio; select color saturation; select color brightness
Step 6	Output Su embroidery effect	Preview effect; save image; upload platform; customize

using MasterSu.

After the experimental task was completed, the subjects were asked to complete a scale based on their authentic user experience. The rating scale draws on the list of questions listed by Kim [5] to validate the sustainable application of digital technology in ICH to assess the solutions provided in this study in terms of education, feasibility, and guidance. In this regard, each dimension corresponds to five questions, all using a Likert five-point scale. Interviews with users were then conducted to understand the users’ feelings about using this system. The usability testing scenario is shown in Figure 13. The usability testing software interface is shown in Figure 14.

5.2. Usability-Test Results

The results for the 10 novice users are shown in

Table 3. Usability test scores.

Dimensionality	Items	Contents	Mean	S.D.
Educational effects	EE1	Does the software effectively simulate the effect of Su embroidery	3.7	0.82
	EE2	After trying to use this software, I have a better understanding of both Su embroidery and Intangible Cultural Heritage	4.2	0.63
	EE3	This software effectively expresses the importance of preserving Su embroidery	3.3	0.95
	EE4	I agree with the idea of preserving Su embroidery conveyed by this software	3.6	1.07
	EE5	I will recommend my friends to try this software	4.1	0.88
Technical effects	TE1	This software is easy to understand and easy to use	3.6	1.07
	TE2	This software made me interested in embroidery and Intangible Cultural Heritage, and I enjoyed participating in it.	4.1	0.88
	TE3	This software helps me to try to make Su embroidery in reality	3.8	1.14
	TE4	Su embroidery is suitable for conservation with digital technology	4.0	0.82
	TE5	This software has more advantages than the traditional learning methods in reality	4.0	1.05
Inducing effects	IE1	This software allows the user to participate in the experience of Su embroidery	3.8	0.79
	IE2	This software can stimulate interest in exploring Su embroidery	3.7	1.06
	IE3	I am going to continue to use this software	4.1	0.74
	IE4	I am going to continue to learn about Su embroidery	4.1	0.99
	IE5	I am going to receive practical training in Su embroidery	3.6	0.97

. From the statistical scores of each option, the mean scores about the educational, feasibility, and guidance evaluations of MasterSu were all greater than 3.5, showing a good evaluation overall, indicating that the subjects who participated in this usability evaluation all thought that MasterSu could promote people’s understanding of Su embroidery to a certain extent and promote the sustainable development of Su embroidery.

6. Discussion

As we can see from the user ratings, MasterSu performs well in terms of education, feasibility, and guidance, validating the effectiveness of the system. Overall, MasterSu performs the best on the feasibility dimension, which means that MasterSu has a better software performance that allows the average person to get started quickly and has some advantages in terms of practical use. The lower scores on the educational and guidance dimensions may be because the performance of these two dimensions needs to be demonstrated through prolonged use, while the feasibility of the software can be directly known through use. However, in general, the subjects’ evaluations of MasterSu were all positive, indicating that the digital solution provided in this paper can introduce more people to the learning of ICH and promote the sustainable development of ICH to a certain extent.

In addition, we analyzed users’ subjective feedback to better understand their feelings about using the software. When evaluating the experience of using the software, the most frequently mentioned keywords were “easy to understand” and “delicate stitches”, indicating that the process of MasterSu is user-friendly for novice users and that more ordinary people can join the learning process of embroidery. When comparing the results generated by the system with those of other software, some users mentioned that “compared with other software, MasterSu guides me step by step in the process of creating embroidery effects and gives me a sense of fun”, which indicates that the images generated by the system are recognized by most users. MasterSu is not a complete replacement for manual work but rather an aid to the user in the creation of embroidery. Of course, MasterSu still has some shortcomings, and some users mentioned some shortcomings of the interaction process and simulation effect, such as “slow loading” and “not enough light effect simulation, not strong enough three-dimensionality”. In view of this problem, we also carried out optimization again.

The MasterSu system can indeed play an innovative role in the design process of Su embroidery, but there are still many difficulties in fully reproducing the intangible cultural skills that have been passed down for thousands of years digitally. The main influence is the diversity and complexity of the stitches used in Su embroidery, which are different for different embroidery objects and expressions. The stitch is created by integrating human thoughts, ideas, and experiences, which includes the flexibility of the actual embroidery process, and this irregularity is the core factor affecting the digital reproduction of the stitch. The complexity of embroidery patterns is also a major constraint to the process of the digital preservation of Su embroidery. Due to the irregularity and complexity of pattern composition, there are still many problems that need to be solved to achieve accurate delineation and conversion of image object contours, and human intervention is still needed to extract image object information. In the future, we will investigate the digitalization of more stitches and the intelligent extraction of image information to expand the functions of the digital-assistant design platform, improve the usability and generalization of the platform, and promote the sustainable development of the digital preservation and inheritance of embroidery.

7. Conclusions and Future Work

This study takes Su embroidery as the research object, analyzes the current development status of Su embroidery, finds the main problems that hinder the sustainable development of Su embroidery, and proposes an automatic Su-embroidery generation system called MasterSu based on the CorelDraw platform. The main contributions of MasterSu are three-fold: (1) MasterSu can automatically generate embroidery manuscripts according to an image, accelerate the production efficiency of Suzhou embroidery, and promote the rapid circulation of Suzhou embroidery in the market. (2) MasterSu allows novice users to participate in the production of Su embroidery by simply dividing the embroidery area and deciding the general direction of the threads, which lowers the entry threshold of Su embroidery and allows more people to understand the culture and production of Su embroidery, thus promoting the sustainable development of Su embroidery. (3) MasterSu can help manufacturers of Su embroidery to reduce the cost of consumer-preference exploration by generating Su embroidery renderings through the Su embroidery computer, which can start consumer preference assessment before the product is produced. Finally, this study also evaluated the MasterSu system through usability testing, and the experimental results showed that this system performed well in terms of education, feasibility, and guidance, confirming the effectiveness of the system in promoting the sustainable development of Su embroidery.

MasterSu will accelerate the innovative design and production efficiency of Su embroidery, promoting its sustainable dissemination and sustainable economic flow. In addition, MasterSu will not only promote the sustainable development of intangible cultural heritage in China, but also provide a digital solution for the sustainable development of intangible cultural heritage worldwide.

However, this study still has several limitations: (1) MasterSu currently only implements digitization methods for the most commonly used sets of stitches in Su embroidery. The system’s stitch library can be further enriched, and the algorithm needs to be continuously optimized in the future. (2) This study has not been able to quantify the amount of work involved in the development of Su embroidery by MasterSu, as the work efficiency of Su embroidery is influenced by the individual embroiderer’s technique and it is not possible to calculate an accurate amount of work. (3) Only usability testing has been used to verify whether MasterSu can promote the sustainable development of Su embroidery. In fact, it takes both time and practice to verify whether a craft can survive the test of time, so the effectiveness of using MasterSu needs to be further observed in the future. The usability test in this study was only conducted by ordinary users because of the limitations of the system, and it will be more conducive to involve more people related to the Su-embroidery heritage, such as developers and ICH inheritors, to test and evaluate the system.