Information integration method of texture and gray based on neural network

From viewpoint of multi-information fusion, the image segmentation method was proposed, in which the information of variation in texture and gray was integrated by means of function-linked neural network. On the basis of the method, the sub-region based region growing algorithm was designed and implemented. First, several sub-regions are assigned, and the texture and gray information of the sub -regions are extracted. Then, the training of the networks is judged and finally the classification for the network is performed to complete image segmentation. The experimental results have shown that the method may obtain good segmentation result for certain images.