Font Size: a A A

The Research And Implementation Of Texture Segmentation Based On Gabor Filters

Posted on:2018-10-29Degree:MasterType:Thesis
Country:ChinaCandidate:P WangFull Text:PDF
GTID:2348330512481398Subject:Engineering
Abstract/Summary:PDF Full Text Request
Image technology is widely used in different fields such as chemical industry,meteorology, water conservancy, medicine and so on. Now the rapid development of image technology not only promotes the productivity development, but also requires engineers to use image, analyze image and comprehend image for servicing the manufacturing,agriculture, service industry. Image segmentation plays an important and fundamental role in image technology, and the research on image segmentation is also increasing year by year. The result of image segmentation determines the understanding of image application, meanwhile will have a crucial impact on further analysis of image.So the image segmentation is very important. Texture segmentation is a difficult problem in the field of image segmentation. On the on hand, the definition of texture was not unified, on the other hand, there are many algorithms for texture segmentation and its effect is not ideal. Therefore, this fact decides the research for texture segmentation is basic and necessary. In this paper, we will focus on the multi texture image segmentation algorithm based on Gabor filters, and use the Matlab software to build a multi-texture image segmentation framework software system.First, from the point of texture segmentation, the paper will summarize the commonly used texture segmentation algorithm. The pros and cons of these algorithms will be tested and explained. Because the algorithm based on Gabor filter was more in line with the cognition of the human visual system, this paper will choose it as research basis. This paper will analyze the thought and principle of Gabor filtering algorithm in detail, and carefully study the parameters of Gabor filters. Meanwhile, we will put forward a selection scheme for according with reality and little computing power through theoretical and literature summary.Second, handling the feature was an important step after extracting feature. This paper analyzes and compares the feature processing and feature clustering. Feature processing is divided into preprocessing and filtering. For feature filtering, this paper selects 4 kinds of filtering algorithms to compare, which the best filtering algorithm be the next process step. For feature clustering, this paper summarizes and compares different clustering algorithm, and selects the K-means algorithm be the clustering methods. Meanwhile, this paper makes a detailed study of the two key points of the K-means methods.For the problem that using Gabor feature for texture was not good, this paper chooses the feature fusion technique to deal with the problem. This paper adopts the idea of equal proportion fusion, which combines the Gabor feature and the location feature. Meanwhile, this paper will combine the Gabor filter and histogram feature. The experimental results show that the algorithm can improve the texture segmentation results.Finally, this paper will use Matlab software to build a multi texture image segmentation software system based on Gabor filters. The system will be analyzed in detail.
Keywords/Search Tags:texture segmentation, Gabor, feature fusion, the location feature
PDF Full Text Request
Related items