A Study on Image Categorization Techniques
International Journal on Orange Technologies, 5 (5):
164-170 (May 2023)Abstract
Image segmentation is the act of splitting a picture into meaningful and non-overlapping parts. It is a crucial step in comprehending natural scenes and has become a hotbed of research in the fields of image processing and computer vision. Even after decades of work and several successes, feature extraction and model design remain difficult. In this article, we carefully review the development in image segmentation techniques. Three crucial stages of image segmentation—classical segmentation, collaborative segmentation, and semantic segmentation based on deep learning—are primarily examined in accordance with segmentation principles and image data characteristics. We compare, contrast, and briefly discuss the benefits and drawbacks of segmentation models as well as their applicability. We also elaborate on the primary algorithms and critical strategies in each stage.