Region Growing and Object Extraction Techniques

For some applications, such as image recognition or compression, whole image cannot be processed 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 image segmentation algorithms into region-based segmentation, data clustering, and edge-base segmentation. Each approach has its own pros and cons. This paper presents a brief review of some of the promising region growing and object-extraction techniques enlisting their advantages and limitations.