Figure 1.
Some interface of our system.
Figure 1.
Some interface of our system.
Figure 2.
The system flow of this work.
Figure 2.
The system flow of this work.
Figure 3.
An example of horizontal composition.
Figure 3.
An example of horizontal composition.
Figure 4.
(a) Original figure. (b) Edge detection results. (c) The longest line and its corresponding angle, as shown near the center of the image.
Figure 4.
(a) Original figure. (b) Edge detection results. (c) The longest line and its corresponding angle, as shown near the center of the image.
Figure 5.
The lines and places (marked in red) where the rule of thirds considers.
Figure 5.
The lines and places (marked in red) where the rule of thirds considers.
Figure 6.
An example image using the rule of thirds.
Figure 6.
An example image using the rule of thirds.
Figure 7.
(a) Original figure. (b) Back-projecting result. (c) Reversed result of (b). (d) Mean shift result of (c). (e) Result of (d) after histogram equalization. (f) The saliency region.
Figure 7.
(a) Original figure. (b) Back-projecting result. (c) Reversed result of (b). (d) Mean shift result of (c). (e) Result of (d) after histogram equalization. (f) The saliency region.
Figure 8.
The main body of the foreground and the center of the foreground.
Figure 8.
The main body of the foreground and the center of the foreground.
Figure 9.
The positions to place the main body of the foreground.
Figure 9.
The positions to place the main body of the foreground.
Figure 10.
An example image following the triangle composition rule.
Figure 10.
An example image following the triangle composition rule.
Figure 11.
The smallest triangle that can contain the foreground.
Figure 11.
The smallest triangle that can contain the foreground.
Figure 12.
An example image using the vanishing point composition.
Figure 12.
An example image using the vanishing point composition.
Figure 13.
(a) Original image. (b) Edge detection result. (c) Vanishing point calculated result (enlarge to see the red detected lines).
Figure 13.
(a) Original image. (b) Edge detection result. (c) Vanishing point calculated result (enlarge to see the red detected lines).
Figure 14.
The reference positions to consider a vanishing point.
Figure 14.
The reference positions to consider a vanishing point.
Figure 15.
An example using the frame within a frame composition rule.
Figure 15.
An example using the frame within a frame composition rule.
Figure 16.
(a) Original image. (b) The GrabCut mask. (c) The obtained background using GrabCut. (d) The resulting frame-within-a-frame foreground.
Figure 16.
(a) Original image. (b) The GrabCut mask. (c) The obtained background using GrabCut. (d) The resulting frame-within-a-frame foreground.
Figure 17.
(a) Original image. (b) Edge detection result.
Figure 17.
(a) Original image. (b) Edge detection result.
Figure 18.
(a) The calculation of integral image. (b) The way to calculate the total pixel values of a rectangular region.
Figure 18.
(a) The calculation of integral image. (b) The way to calculate the total pixel values of a rectangular region.
Figure 19.
(a) An example image with (horizontal) balance of intensity. (b) An image without balance of intensity.
Figure 19.
(a) An example image with (horizontal) balance of intensity. (b) An image without balance of intensity.
Figure 20.
(a) Original image. (b) The intensity distribution for left half image. (c) The intensity distribution for the right half image.
Figure 20.
(a) Original image. (b) The intensity distribution for left half image. (c) The intensity distribution for the right half image.
Figure 21.
(a) An example image with high contrast. (b) An image with low contrast.
Figure 21.
(a) An example image with high contrast. (b) An image with low contrast.
Figure 22.
(a) The distribution when rule of thirds is considered. (b) The distribution for a random photo taking process.
Figure 22.
(a) The distribution when rule of thirds is considered. (b) The distribution for a random photo taking process.
Figure 23.
(a) The distribution when the focus composition rule is considered. (b) The distribution for a random photo taking process.
Figure 23.
(a) The distribution when the focus composition rule is considered. (b) The distribution for a random photo taking process.
Figure 24.
(a–d) Original images. (e–h) Foregrounds of (a–d), respectively.
Figure 24.
(a–d) Original images. (e–h) Foregrounds of (a–d), respectively.
Figure 25.
(a) Main interface. (b) The score for each composition method. (c) The interface for setting/selecting composition method (enlarge to see more details).
Figure 25.
(a) Main interface. (b) The score for each composition method. (c) The interface for setting/selecting composition method (enlarge to see more details).
Figure 26.
(a) Explanation for a composition method. (b) The setting of a score value, and when the evaluation score is equal to or greater than this score, the photo will be automatically taken, and the interface will be changed to (c).
Figure 26.
(a) Explanation for a composition method. (b) The setting of a score value, and when the evaluation score is equal to or greater than this score, the photo will be automatically taken, and the interface will be changed to (c).
Figure 27.
First testing image.
Figure 27.
First testing image.
Figure 28.
(a) Horizontal composition. (b) Rule of thirds composition. (c) Triangle composition. (d) Focus composition.
Figure 28.
(a) Horizontal composition. (b) Rule of thirds composition. (c) Triangle composition. (d) Focus composition.
Figure 29.
Second testing image.
Figure 29.
Second testing image.
Figure 30.
(a) Central composition. (b) Triangle composition. (c) Vanishing point composition (enlarge to see the red detected lines). (d) Focus composition (enlarge to see the detected focal point marked in yellow around the upper right region).
Figure 30.
(a) Central composition. (b) Triangle composition. (c) Vanishing point composition (enlarge to see the red detected lines). (d) Focus composition (enlarge to see the detected focal point marked in yellow around the upper right region).
Figure 31.
Third testing image.
Figure 31.
Third testing image.
Figure 32.
(a) Horizontal composition. (b) Vanishing point composition. (c) Frame within a frame composition. (d) Focus composition.
Figure 32.
(a) Horizontal composition. (b) Vanishing point composition. (c) Frame within a frame composition. (d) Focus composition.
Figure 33.
The scores for the first scene using the composition methods of horizontal, rule of thirds, triangle, and focus.
Figure 33.
The scores for the first scene using the composition methods of horizontal, rule of thirds, triangle, and focus.
Figure 34.
The scores for the second scene using the composition methods of horizontal, vanishing point, frame within a frame, and focus.
Figure 34.
The scores for the second scene using the composition methods of horizontal, vanishing point, frame within a frame, and focus.
Figure 35.
The limitation of finding the vanishing point when the input image is like (b).
Figure 35.
The limitation of finding the vanishing point when the input image is like (b).
Figure 36.
The result of extracting the foreground in a frame-within-a-frame composition method when a closed frame is encountered.
Figure 36.
The result of extracting the foreground in a frame-within-a-frame composition method when a closed frame is encountered.
Figure 37.
The result of extracting the foreground in a frame within a frame composition method when an open frame is encountered.
Figure 37.
The result of extracting the foreground in a frame within a frame composition method when an open frame is encountered.
Figure 38.
Results of extracting the foreground in a frame-within-a-frame composition method using different resolutions. (a) Original image, with the resolution of 1200 × 1600. (b) The extracted foreground result using (a). (c) The extracted foreground result using a reduced resolution of 171 × 228.
Figure 38.
Results of extracting the foreground in a frame-within-a-frame composition method using different resolutions. (a) Original image, with the resolution of 1200 × 1600. (b) The extracted foreground result using (a). (c) The extracted foreground result using a reduced resolution of 171 × 228.
Figure 39.
Results of extracting the foreground in a frame-within-a-frame composition method using different resolutions. (a) Original image, with the resolution of 4032 × 3024. (b) The extracted foreground result using (a). (c) The extracted foreground result using a reduced resolution of 201 × 151.
Figure 39.
Results of extracting the foreground in a frame-within-a-frame composition method using different resolutions. (a) Original image, with the resolution of 4032 × 3024. (b) The extracted foreground result using (a). (c) The extracted foreground result using a reduced resolution of 201 × 151.
Figure 40.
Results for the horizontal composition method.
Figure 40.
Results for the horizontal composition method.
Figure 41.
Results for the rule-of-thirds composition method.
Figure 41.
Results for the rule-of-thirds composition method.
Figure 42.
Results for the triangle composition method.
Figure 42.
Results for the triangle composition method.
Figure 43.
Results for the vanishing-point composition method.
Figure 43.
Results for the vanishing-point composition method.
Figure 44.
Results for the focus composition method.
Figure 44.
Results for the focus composition method.
Figure 45.
Results for the frame-within-a-frame composition method.
Figure 45.
Results for the frame-within-a-frame composition method.
Figure 46.
Results for the intensity balance composition method.
Figure 46.
Results for the intensity balance composition method.
Figure 47.
Results for the contrast composition method.
Figure 47.
Results for the contrast composition method.
Figure 48.
An image with good contrast.
Figure 48.
An image with good contrast.
Figure 49.
The result of preference.
Figure 49.
The result of preference.
Figure 50.
Comparison of average score between with/without using our system.
Figure 50.
Comparison of average score between with/without using our system.
Table 1.
The execution time for frame within a frame, in terms of seconds.
Table 1.
The execution time for frame within a frame, in terms of seconds.
Resolution | Time |
---|
960 × 720 | 5.658 |
320 × 240 | 0.207 |
192 × 144 | 0.073 |
Table 2.
System environment.
Table 2.
System environment.
Operating System | Windows 10 64bit |
Processor | Intel(R) Core(TM)i7-6700 CPU 3.40GHz |
Display card | NVIDIA GeForce GTX 960 |
Memory | 32GB |
Development Platform | Android Studio, Eclipse |
Mobile Device | Sony Xperia Z2 |
Programming Language | Java |
Library | OpenCV-3.4.1, Jama |
Table 3.
The score for each composition method in the first testing image.
Table 3.
The score for each composition method in the first testing image.
Composition Method | Score | Composition Method | Score |
---|
horizontal | 100 | rule of thirds | 98 |
triangle | 91 | focus | 75 |
intensity (90) | 76 | contrast (50) | 98 |
Table 4.
The score for each composition method in the second testing image.
Table 4.
The score for each composition method in the second testing image.
Composition Method | Score | Composition Method | Score |
---|
central | 98 | triangle | 89 |
vanishing | 97 | focus | 47 |
intensity (90) | 91 | contrast (50) | 95 |
Table 5.
The score for each composition method in the third testing image.
Table 5.
The score for each composition method in the third testing image.
Composition Method | Score | Composition Method | Score |
---|
Horizontal | 100 | vanishing | 91 |
Frame within a Frame | 71 | focus | 83 |
intensity (90) | 94 | contrast (50) | 96 |