Particle Swarm Optimization Algorithm For Image Segmentation In Mixed-size Foam

Froth flotation image is researched in this paper, mainly the segmentation problem of it. Characteristic of flotation froth image is analyzed combining with flotation craft parameters. On the base of introducing dynamic and static characteristics of the froth image, morphology features extraction method of the bubble image is studied. Moreover, reasonable method to extract morphological features of bubble image is proposed. It lays the foundation for the optimal control of mineral flotation system which is based on digital image processing.First of all, aiming at the characteristics of bubble image and it is vulnerable to impacted by light condition, it is difficult to effectively separate a variety of sizes of bubbles in froth image. At the same time, the theory and parameter setting of particle swarm optimization, threshold segmentation algorithm of the bubble image and its deficiencies are introduced.Second, on the basis of analyzing the features of bubble image deeply, initialization segmentation algorithm based on the bottom edge of the bubble image segmentation is proposed. And improved particle swarm optimization is used to improve the edge threshold accuracy of the bottom edge segmentation. Through froth image initialization segmentation, there are still over-segmentation and under segmentation in the bubble images. Then segmentation by means of region merging and splitting is proposed.Finally, several evaluation criteria of image segmentation are analyzed. And then bubble image segmentation algorithm which is presented in this thesis is compared with other segmentation algorithms. Comparison result verifies the effectiveness and accuracy of this algorithm.The bubble image feature extraction method in this thesis has been applied in the flotation froth video monitoring system. The method has high stability and robustness. It can accurately extract morphological features of froth flotation under different conditions.