Study On The Recognition Method Of Halftoning Image Classification

The pre-knowledge is important to inverse halftoning,so researching the recognition algorithm of halftoning image classification will improve the performance of inverse halftoning methods on the self-adaptation,optimization and establishment.Having studied the principle of different halftoning methods,the texture property of halftone image is analyzed.By researching the existing analysis approaches of image texture,it is found the texture eigenvectors of the two gradient preponderances produced by Gray Level-gradient Co-occurrence Matrix and the distributing proportion of wavelet energy can be used for describing the texture feature of halftone image. Finally,a recognition method of halftoning image classification is proposed by two gradient preponderances,the distributing proportion feature of wavelet energy and Support Vector Machine.The three kinds of kernel function for SVM are simulated and prepared on their performances.Experimental results show that the correct recognition rate of Radial Basis Function is higher than Multinomial Function and Sigmoid Function.