In order to deal with multi-object image effectively, the paper used grey relation grade to measure the similarity berntween different pixels, and transformed the normalized cut criterion into the dual-threshold segmentation criterion, then optirnmized the criterion using particle swarm optimization algorithm to get the best thresholds. Experiments show that the method rnhas less segmenting time and can segment the multi-object image accurately.%为了对多目标图像进行有效分割,使用灰色关联分析来衡量像素点间的相似性,并将normalized cut准则推导为双闽值分割准则,最后使用粒子群算法优化该准则,得出最佳的分割阈值.实验表明该方法计算速度快,能有效分割多目标图像.
展开▼