Image Segmentation And Classification Based On Cat Swarm Optimization

Image segmentation and image classification are both important research in the field of image processing and computer vision. These technologies are effective or not decisive influence the follow-up work of image processing and computer vision. With the diversity and large of the images, the traditional segmentation algorithms and classification algorithms have some limitations, so more and more new algorithms are used to study the problems of image segmentation and image classification, the method of image segmentation and classification based on swarm intelligence algorithm is an important direction. However, the classical swarm intelligence algorithms also have the shortcomings, such as slower, easy to fall into local optimum, the complexity of the larger, etc. Aiming at these problems above, cat swarm optimization algorithm is proposed to study the problem of image segmentation and image classification. In this paper, we analyze cat swarm optimization algorithm, and its improvement and application The main research work is as follows:1) A new hybrid median filtering algorithm is proposed. Aiming at the fact that de-noising effect of median filtering and hybrid median filtering deal with the big probability noisy image is insufficient, we introduced the adaptive thought based on the original hybrid median filter algorithm so that the window size can be adjusted by the hybrid median algorithm adaptively and then achieve better de-noising effect. Experimental results show that the new filtering algorithm has better performance in eliminating noise compared with the original algorithm.2) An image segmentation method based on nearest neighbor clustering of cat swarm optimization algorithm is proposed. According to principles and steps of cat swarm optimization algorithm, the thought of the nearest neighbor is introduced to cat swarm optimization algorithm so as to cluster. Finally, we perform image segmentation experiments by natural image and structure image in the standard image library using the combined clustering algorithm. The validity and good segmentation effect of this method is verified by experiments.3) An image objects classification method based on improved cat swarm optimization algorithm is proposed. Aiming at the problem that time consumed by cat swarm optimization algorithm in processing large data images is very long, a new dynamic inertia weight is introduced to the tracking model of cat swarm optimization algorithm. We change the constant acceleration coefficient into a variable one simultaneously, and make it change dramatically as the circumstances require. Finally, the improved cat swarm optimization algorithm is used in experiments of objects classification. The practical effectiveness of this method is verified by the experimental results, and these results also show that running time of the improved cat swarm optimization algorithm is shorter than the original cat swarm optimization algorithm and speed the improved algorithm increased as well.4) An system of image segmentation and classification based on cat swarm optimization algorithm is designed. Combined with all previous work, a system interface which facilities user operation and usage is designed, this system includes image restoration module, image segmentation module and image classification module. The results show that this system can operate normally, and has good stability and practicability.