Semantic Image Segmentation using Canny-Edge Detector

Semantic Image Segmentation (SIM) is one of the most important areas of image processing. It is used in different fields like robotics, satellite imaging, Biometric verification etc. Image segmentation system depends mainly on two processes, i.e. image feature extraction and object class learning. Some of the researchers have exploited image feature like Texton-Feature, by the help of Texton feature edge, color and position of an object can be detected precisely. Due to this extraordinary quality of texton, we use it for features extraction to get accurate results about objects. Both features extraction and object class learning are essential for precise segmentation of objects. In this work, we have only focused on feature extraction phase to provide best results comparative to all state-of-the-art works. Therefore, we have applied some processes to the image to improve the feature extraction phase. We used Canny-Edge filters/detector (C-EF) to detect the potential shapes and edges parameters for objects. Using C-EF all the results we have achieved are quite satisfactory and have outstep state-of-the-art work.