多蜜蜂重叠图像的凹点匹配分割算法  

Multi-bee overlapping segmentation algorithm based on the pits matching

作　　者：白云飞 杨梦磊 张翔[1] 林建[1] Bai Yunfei;Yang Menglei;Zhang Xiang;Lin Jian(College of Information Engineering,Fujian Agriculture and Forestry University,Fuzhou,350002,China)

机构地区：[1]福建农林大学机电工程学院,福州市350002

出　　处：《中国农机化学报》2020年第6期188-193,共6页Journal of Chinese Agricultural Mechanization

基　　金：福建省高原学科建设项目(712018014)。

摘　　要：针对蜜蜂在巢脾上以不同方式发生重叠的情况,提出一种先匹配凹陷区域再识别对应凹点对的分割方法。考虑到蜜蜂的颜色特征,在HSV空间内,利用OSTU算法对V分量处理得到分割阙值T,将蜜蜂个体从巢脾背景中分割出来。根据得到的蜜蜂个体图像特征,利用形态学方法除去足和触角,获得仅含躯干部分的重叠二值图像。使用孔洞填充和小面积去除法除去孔洞和噪声。根据凹点数和蜜蜂的数量关系,判断蜜蜂重叠状态。串联条件下,先进行内外凹区域的匹配,再搜寻对应凹点对,完成重叠蜜蜂的分割。并联条件下,由于不存在内凹陷区域,不用进行凹陷区域匹配,求取重叠区域中心坐标与各个凹陷区域上点的欧氏距离最小值,最小值对应的点即为到凹点。在MATLAB环境下,对由74只蜜蜂个体组成的含有6种重叠状态的16组重叠个体进行分离试验,分离重叠个体的成功率达到93.75%。利用凹点匹配在单目视觉下可以实现对多重叠状态的蜜蜂个体进行分离的目的。In view of the different ways of the honeybee overlapping on the nest spleen,a new method is proposed to match the sunken area and then identify the corresponding concave pairs.Taking the color characteristics of bees into account,in the HSV space,the segmentation threshold T is obtained by using the OSTU algorithm for V component,and the individual bee is segmented from the background of the nest spleen.According to the features of the individual bee,the feet and antennae were removed by morphological method,and the overlapping binary images containing only the trunk were obtained.Hole filling and small area removal methods are used to remove holes and noise.The overlapping state of bees is judged according to the relationship between the number of concave points and the number of bees.Under the condition of series,the internal and external concave areas are matched first,then the corresponding concave pairs are searched to complete the segmentation of the overlapping bees.The minimum Euclidean distance between the center coordinates of the overlapping region and the points on each depression region is obtained,and the corresponding point of the minimum value is the concave point.In matlab environment,16 groups of overlapping individuals with 6 overlapping states composed of 74 bee individuals were tested for separation,then the success rate of separating overlapping individuals reached 93.75%.The purpose of separation of honeybee individuals with multiple overlapping states can be achieved by using concave matching in monocular vision.

关 键 词：重叠蜜蜂 匹配原则 凹点识别 串并联 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]