The Research Of The Gabor Filter In The Texture Analysis Applications

The texture is a very important visual property of the object surface, the surface of different objects will generate different texture, it playes an important role in object recognition and image analysis. For example, in remote sensing image analysis, document processing, image retrieval, industrial surface inspection and other fields so on, the texture analysis play a very good role.So far, there are many methods of the analysis about the texture. We can divide them into four categories:statistics-based, structure-based, model-based, and signal processing method. Now, texture analysis based on Gabor filter is a very popular way. Under the uncertainty principle, Gabor filter is the only one function which can obtain the best information in both spatial domain and frequency domain. And the2D-Gabor, which is proposed to solve the problem of picture processing, is familiar with the biological visual system, that’s why people love the Gabor so much.In the article, we study about the texture and its characteristics firstly, and after understanding the advantages of Gabor, we did a lot of research about the application of the Gabor. We did a lot of research about the two field in which the Gabor filter play a very important role, such as fingerprint image enhancement and texture image segmentation.. First we did a lot of work about the usage of Gabor on fingerprint enhancement, after understanding the essence of the classic way based on Gabor, we try to make some progress in the fingerprint enhancement, and at last we did it, the improved Gabor filter cost less time than before. Then we did a lot of research about the feature extraction of texture image, which is based on Gabor filter. Because the Gabor-wavelet is not good enough to get the high-frequency characteristics of an image, after a lot of work, we find the Log_Gabor filter is a very good choice to overcome the weakness of Gabor-wavelet. So far, the Log_Gabor filter is well used on the palmprint and iris recognition, in this article we use it to extract the feature of an texture image and combined with the K-means clustering method, we finished the segmentation of an texture image.