Image Segmentation Based Transition Region

Image Segmentation is the first step of Image Recognition, whether the segmentation is correct or not will affect the later processing directly and it is a key problem in computer vision. In this paper we mainly studied the direct transition region extraction(D-TRE) based traditional image segmentation based transition region extraction .By analyzing the drawbacks of the traditional transition region extraction methods we give a more generalized definition on transition region, and suggest that the searching aspects of TRE should be converted to study methods that can make transition region directly extracted. Besides we give clear explanation of the thresholding criteria geometrically and mathematically.In gradient based TRE method study we present degree based TRE and image segmentation method(D-TREM) and wavelet energy ratio based TRE and segmentation method(W-TREM). In DTREM we introduce the information theory into the transition region extraction, this method have a good performance to deal with salt and pepper noise, but the speed is comparatively slow. By introducing the wavelet into the transition region extraction and disposing the image using the 2_D discrete wavelet transition we computer the wavelet energy and using this ratio segment the image. WJTREM is only better for segmenting special type of texture images than for other kind of images. D_TREM is better than other gradient-based TRE method.We studied the direct of non-gradient method and deeply discussing local entropy based TRE and image segmentation method (LE-TREM) and local complexity based TRE and image segmentation (C-TREM). LE-TREM introduces the information theory into the transition region extraction and improves the performance to deal with salt and pepper noise. C-TREM is the simplification of the LE-TREM and this method not only has a good performance to deal with salt and pepper noise but also has a fast speed and not has the problem of small sample, and is a method that worth spreading. Both methods have a good performance to deal with salt and pepper noise, which comes from their way measuring grayscale level information. Among the processing of these methods, we using many remote sensening images. Through these study and compare we try to provide a new and effective method in remote sensening image segmentation .