The Study On Lymphoid Tissue Color Pathological Images Synergetic Classification Based On The Textural Feature Space

The lymphoid tissue color pathological images are very complicated, and the different images are very similar in vision, so it is difficult to distinguish them by naked eye. As the science and technology developing, we can analyze the image and calculate the data with computer to help the doctors. The purpose of this paper is to design a fast and accurate classification for lymphoid tissue color pathological images. According to the limitation of the traditional pattern recognition system, this paper proposes a new classification for lymphoid image combining the traditional pattern recognition methods and synergetic neural network. The images this paper processes are got under 100 times microscope, and the different images can be distinguished with textural features. So this paper extract the textural features of image, then select the useful features automatically, lastly classify the images with synergetic neural network. The main content and researching result are follows: Firstly, according to the texture characteristic of the lymphoid tissue color pathological image, we use the texture spectrum, the Fourier Transformation and laws texture measure to extract the image textural features from different points of view. Then, we use the value reduce method in Rough Set to select the useful features for following classification ,at the same time ,we set the features weight combining the subjective weight and objective weight . Secondly, we study the synergetic neural network and propose several solving method according to the limitation of the key technologies in the synergetic neural network. For selecting of prototype patterns, we propose a cluster algorithm based on Rough Set. For setting of attention parameters, we propose a new way with image similarity. Finally, we propose a synergetic classification of lymphoid tissue color pathological images based on the textural features space. We use the textural features instead of the gray features to classify the lymphoid tissue color pathological images in the synergetic neural network. And according to the problems of combing the texture and synergetic neural network, we propose the solving methods. At last, we give the experiment result.