Research Of Image Texture Feature Extraction

In recent years, along with information multimedia time arrival, as well as network inworldwide scale day by day popular, cloud computation being in fashion, the people are moreand more bigger in the daily life work contact information content. The image took theinformation one kind of carrier, has, the information content intuitively big, is advantageous forthe characteristic which between the different country exchanges, is the network multimediaimportant constituent. Based on the text image retrieval is based on the content image retrievalfoundation, with the artificial way explanation pictorial information, its work load we imagineswith difficulty, the feasibility is also worth discussing. Therefore CBIR method effectiveaddressing this difficult problem . Based on content image retrieval (CBIR) including four stages,respectively is: Gain image, extraction characteristic, classified image, retrieval image. Theimage retrieval mainly is a core question: Which one kind of image characteristic selects whatalgorithm to withdraw, fast effective carries on the image the discrimination and the examination.The texture characteristic extraction is one of CBIR key question, the present paper also is basedon the image texture characteristic extraction is a foundation.Although until now, already had the very many method extraction image texturecharacteristic, and might carry on the classification to the texture, the analysis characteristic, butwe could not choose a best method to carry on the image texture characteristic to indicate,thought in the actual application, they had the merit and the shortcoming respectively, and manyalgorithms if did not pass through the improvement, the existence computation load was big, thestorage space took lacks the practical application value excessively much the shortcoming.Therefore, will improve the original algorithm as well as carries on many kinds of algorithmcombination still was a later texture analysis important direction. Based on the method ofaverage foundation, the present paper has carried on the analysis and the improvement to GLCM,proposed one kind of fast gradation - gradient GLCM and one gradation neighborhood elementGLCM. Based on the wavelet transformation, the present paper proposed the wavelettransformation improvement algorithm, carries on the image texture characteristic extraction, hasintroduced the Refined parameters and the logarithm - polar coordinate transformation method.