Recent Techniques in Image Feature Extraction

Dear Colleagues,

In the modern era of deep learning and foundation models, feature extraction still remains a cornerstone of computer vision and image processing, facilitating tasks such as object detection, class activation map generation, face recognition, or transfer learning for semantic segmentation and depth prediction. By transforming raw image data into representations with meaningful semantics, it simplifies data handling, improves model accuracy, and filters out noise. Feature extraction also ensures consistency under varying conditions, supporting efficient image retrieval and advanced applications (such as autonomous driving and medical diagnostics). It also facilitates multimodal data analysis, enriching data representation and boosting overall model performance.

Self-supervised large-scale feature extraction techniques are now ubiquitous, enabling tasks to be tackled in few-shot or even zero-shot regimes. However, several challenges remain, reflecting the complexity and dynamic nature of the field. Handling high-dimensional data and ensuring robustness to variations in illumination, scale, and occlusion are major hurdles. Noise and artefacts in images complicate the extraction process, especially in real-time applications that require fast efficient techniques. Deep features often need to be up-sampled back to the original input resolution, which can introduce unwanted artefacts into the output map if carried out naively.

Integrating traditional feature extraction with deep learning approaches and understanding their inner workings remains complicated. As models become more complex, understanding and interpreting the extracted features and their impact on the final decision process becomes essential, especially in critical applications that rely on trustworthy and explainable systems.

The aim of this Special Issue is to showcase novel works of research, as well as review papers on state-of-the-art techniques related to feature extraction and their applications, highlighting the ongoing evolution and future directions in this critical area of research. These advances will drive further innovations in various domains that rely on effective image feature extraction.

Prof. Dr. Anastasios L. Kesidis
Dr. Giorgos Sfikas
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Journal of Imaging is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1800 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• feature extraction
• image processing
• feature selection
• machine learning
• computer vision
• deep features
• self-supervised learning
• transformers
• multi-scale features
• graph-based feature extraction
• attention mechanisms
• contrastive learning
• feature upsampling