Advances and Insights in Image Texture Analysis : A Review

Abstract

Texture analysis is an essential step in image analysis, and subsequent applications such as medical imaging, remote sensing, and scene understanding are highly important in image processing. Although it is vital, the field presents its own set of research challenges, especially when manipulating variations in texture patterns and requirements for properties that remain unaffected by related transformations, such as rotation, scaling, and translation. This review provides an in-depth description of key activities in the field of texture analysis, including classification, segmentation, synthesis, and image retrieval, along with their strengths and limitations. The approaches are classified as structural, statistical, and model-based and are discussed in consideration of their appropriateness and performance. The regular textures most favour structural methods, whereas statistical and model-based methods are more flexible, although they sometimes require more computational resources. Other challenges that are outlined in the review include a lack of support for real-time and transform-invariant applications. These findings can aid in determining the appropriate techniques to use and in developing lightweight yet durable methods for texture analysis. Overall, the review offers profound insights into the field and provides a course for future research and creativity.