摘要
提出一种自然场景下对不同形状植物果实图像进行识别提取的方法。采用RGB色差模型及Ostu自适应阈值分割进行图像二值化。应用数学形态学方法对二值图像进行预处理获取植物果实轮廓。采用果实质心、最小凸边形方法对目标进行特征提取。实验表明,除对类圆果实识别较好外,对不规则果实识别效果更佳,其最小凸边形提取比外接矩形及类圆提取方法更接近真实图像,具有一定的通用性。
A recognition method of difformity mature fruits is presented . Firstly, the color image was transferred into binary image using RGB chromatism and Ostu adaptive threshold segmentation, then the fruits rough sketch was extracted through applying mathematical morphology, lastly, the object feature was extracted by labeled the centroid and marked fruit edge using minimum convex edge. Experience show： the method is adapted to difformity mature fruits, especially irregular shape fruits, and the feature extracted of convex edge is better than circum- scribed rectangle and semblable circle.
出处
《科学技术与工程》
北大核心
2012年第21期5135-5138,5143,共5页
Science Technology and Engineering
基金
东华理工大学校长基金项目(DHXK1107)资助
关键词
植物果实
图像识别
色差模型
最小凸边形
plant fruits image recognition chromatism model minimum convex edge
