Research On Complex Texture Image Segmentation Based On Local Operator

It is an important task and fundamental issue to study texture image segmentation in the field of image processing.Due to the complex structures,various features and shapes,and the immature cognition of human,texture image segmentation is still a difficult and hotspot problem in the field of image processing.The texture segmentation methods can be generally made up of two processes: feature extraction and segmentation.Due to the diversity and complexity of textures,it is difficult to describe texture features accurately by using the existed texture feature extraction operators.After feature extraction,the segmentation model is used to complete the texture image segmentation combined with the texture features.Owing to the completed closed curve,completed mathematical theory support and topological invariance,the level set method has already shown good performance in the field of image segmentation.In this dissertation,we propose the new local operators to extract texture features and then combine the level set method to segment texture images.The main works of this dissertation are summarized as follows:(1)We firstly introduce and summarize the study significance.And then,the traditional methods and the existed difficulties of texture image segmentation are analyzed.Specially,we summarize the local operators and its application in texture image segmentation.Meanwhile,several common texture image segmentation techniques are given and analyzed.(2)A multi-feature based level set method is proposed to segment texture images.Specifically,two novel local operators named local connection operator and local difference operator are proposed.Then the intensity information and texture features extracted by the two proposed operators are integrated into the level set method to obtain the final segmentation results.The local connection operator is used to describe the number of points which have similar intensity value with the center pixel in the local region.The local difference operator is used to describe the intensity differences between the local region pixels and the center pixels.Finally,the effectiveness and advantages of the proposed method are proved through a series of experiments.(3)A texture segmentation method is proposed by using Gabor filter and improved LTP operator to extract texture features.Specifically,Gabor filter is used to extract the similar texture features;the LTP operator is improved to accurately describe the local feature.Finally,the Gabor filter and the improved LTP operator are integrated into the level set method to obtain the final segmentation results of texture images.The effectiveness and advantages of the method are testified by making lots of experiments.