Study On Technology Of Texture Feature Fxaction Based On Wavelet Transform

As a basic feature of image, texture is a pretty important information source in image resolution. Texture analysis technology has been widely used in computer version, mode identification and image treatment. At the basis of clarifying the texture analysis and depicting the measures, problems about the texture classification which has been a main branch in texture analysis technology are pointed out.Firstly, texture feature extraction which is crucial in this field is introduced, and the traditional texture feature extraction arithmetic is overviewed. The area of application and the effects of the traditional measures are analyzed. Aim at the current image resolution continual enhancement, the shortcomings of the traditional measures in tackling with new problems and the advantages that signal arithmetic has compared to traditional measures are pointed out. Dual-Tree Complex Wavelet Transform (DTCWT) has deepened and extended the Wavelet Transform. It includes two parallel filters which delay is a sampling value interval. So this can get the sample value which can't be got because of space sampling. Odd-Even filter which used alternately during the different levels can presented pretty symmetries. The DTCWT is extended to two dimension image, as a result, it can obtain the direction selection and pretty high redundancy value image wavelet texture feature.Then, study the classification .Pointed out that the concept of knowledge, general classification algorithms and support vector machine classifier. Through the experiment, DTCWT transform of the statistical characteristics of the sample parameters as classification, applied to the support vector machine (SVM). Then,it can be a relatively accurate classification results. This method to extract the algorithm can be directly applied to the UAV radar.