Tutorial: Image Segmentation

For some applications, such as image recognition or compression, we cannot process the whole image directly for the reason that it is inefficient and unpractical. Therefore, several image segmentation algorithms were proposed to segment an image before recognition or compression. Image segmentation is to classify or cluster an image into several parts (regions) according to the feature of image, for example, the pixel value or the frequency response. Up to now, lots of image segmentation algorithms exist and be extensively applied in science and daily life. According to their segmentation method, we can approximately categorize them into region-based segmentation, data clustering, and edge-base segmentation. In this tutorial, we survey several popular image segmentation algorithms, discuss their specialties, and show their segmentation results. Moreover, some segmentation applications are described in the end.