Image segmentation algorithm for local region characteristics through nonsubsampled contourlet transform

Abstract

An image segmentation algorithm for local region characteristics through nonsubsampled contourlet transform comprises the following steps: firstly the nonsubsampled contourlet transform is carried out on an image; then the expression abilities of a local extreme value and local standard deviation on the edge of the image are utilized, and the local standard deviation and the local extreme value on each sub-band are extracted to be used as characteristic vectors; and the characteristic vectors are classified by using FCM, and accordingly the image is segmented. The method provided by the invention can be used for effectively segmenting texture images of multiple combinations, and has an excellent segmentation effect and superior segmentation performance. According to the algorithm, multiscale segmentation is taken into account, and image local region statistical characteristics are also taken into account, so the algorithm is a promising image segmentation technology.

Description

Technical field [0001] The invention relates to a texture image segmentation algorithm, in particular to an image segmentation algorithm of non-subsampling Contourlet transformation of local area features. Background technique [0002] Image segmentation has a wide range of applications in computer vision, object recognition, etc., and the quality of the segmentation has a very important impact on the subsequent processing results. The segmentation method of the image is also different due to the different methods applied. In recent years, with the popularity of wavelet transform, the application of multi-scale segmentation methods based on transform domain is increasing. For example, Liu Guoying and others used the wavelet domain hierarchical Markov model to realize texture image segmentation in Document 1 "Texture Segmentation Based on Wavelet Domain Hierarchical Markov Model" (Journal of Wuhan University·Information Science Edition, 2009, 34(5)). Although the wavelet tr...

AI Technical Summary

Problems solved by technology

However, multi-scale segmentation is often difficult to obtain accurate segmentation results due to insufficient statistical information, especially insufficient local statistical information, which will make it difficult for the final segmentation results to form large and consistent texture regions.

Images