Texture Classification Using SVM And The Fast Algorithm Of Two-dimensional Wavelet Transform

The classification of texture image in biomedical engineering, remote sens-ing telemetry, industrial product testing and other fields has important ap-plication value. Thirty years, the classification of texture image has been popular with attention, puts forward all kinds of texture feature extraction, classification and methods, but most current classification algorithm universal existence many defects (computational complexity, correct classification rate Low), to some extent, restrict the use of the algorithm, and how these extract effective texture feature and accurately classification is also needs further re-search and exploration, some new theories, new technology continuously put forward for further improvement of texture classification algorithm provides the broad prospect.Double tree Kingsbury puts forward the wavelet transform complex field in recent years is the hot research direction, emerging complex-wavelet trans-form has six direction, its advantage is the invariant with approximate trans-lation and rotation invariant with similar Gabor transform, and the charac-teristics and have fewer redundancy, is texture feature extraction very good method.This article tried to put forward a kind of texture classification method, using two-dimensional wavelet transform the fast algorithm for feature extrac-tion, wavelet transform each band outputl1 norm as texture classification, and according to the characteristics of scattering characteristics itself, and then the weighted classified by using support vector machine. From the experiment can see the method in this paper and complex wavelet method has been a basic classification of similar effect, but time is obviously less than a classification of complex wavelet method.