自然场景下成熟苹果彩色图像分割方法的研究  

Research on Mature Apples′ Color Image Segmentation Methods in Natural Outdoors

作　　者：牛希泉[1] 梁艳梅[1] 

机构地区：[1]南开大学现代光学研究所

出　　处：《光电子．激光》2007年第12期1453-1456,共4页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金资助项目(60677012);南开大学学生创新基金资助项目

摘　　要：提出一种在RGB颜色空间分割彩色图像的新方法。针对自然场景下成熟苹果图像,首先通过在RGB颜色模型中寻找灰色分割面来得到最优颜色函数(OCF),将多维空间中的分割问题转化到一维空间,根据目标和背景的不同特点,提出了两种可供选择的获取OCF的方案;然后通过Fisher评价函数找到最佳阈值;最后用数学形态学滤波除去高频噪声。该方法可以完全避免人为干预,而且分割速度快。实验结果表明,该方法可以把自然场景中的苹果有效地分割出来。A new color image segmentation method in the RGB color space is reported in e paper, Firstly, the optimum color functions（OCF） ,to study color images of mature apples in natural outdors,are calculated by looking for the gray segmentation plane in the RGB color model, and the segmentation problem in multidimensional space is translated into one dimension space. Two alternative schemes of acquisition of OCF are presented on the basis of different characteristics of target and background. Then,the optimum threshold is obtained based on the Fisher criterion function. Lastly, high-frequency noise is removed by mathematical morphology. The merits of this segmentation method are that it can avoid human interference and has high efficient segmentation. Experimental results show that the method can segment apples from natural outdoors effectively.

关 键 词：彩色图像分割 FISHER评价函数 数学形态学 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]