Classification And Recognition For Halftoning Image

Different halftoning images need different methods in the application. It is very important for halftoning images to classify and identify before the processes such as inverse halftoning and compression.Firstly, the currently neural net algorithm of halftoning image classification presented by Pao-Chi Chang is studied. And it is simulated using many kinds of halftoning images. But the result can not meet the needs because the features in the algorithm have some limitations. The texture features based on the gray level co-occurrence matrix is selected to improve the algorithm. Experimental results show that the precision can be improved.The complexity of the method above is very high, so a new algorithm is proposed. It used the texture features based on the property of human visual perception to perform the classification and recognition. These features are selected from the gray run-length matrix. The result of classification and recognition is improved, and the complexity is reduced.