套袋苹果影像的分割方法研究

Study on Image Segmentation for Bagged Apple

为了准确识别并分割套袋苹果,对影像中的套袋苹果及其背景在RGB颜色空间模式下,分析各目标物的R(红)、G(绿)及B(蓝)值分布特征,根据各目标物所呈现的R、G及B值差异选取适合于套袋苹果的分割条件,对套袋苹果进行分割,将分割结果与多阈值分割法以及K均值聚类法分割结果进行比较。结果表明:分割法优于以上2种方法,同时,根据该试验提出的分割方法计算苹果识别率,识别率可达93.4%,表明该试验算法对套袋苹果的分割准确率高。

In order to accurately segment of bagged apple,the feature of R,G and B values of RGB mode for bagged apple and other targets image was analyzed,and then the bagged apples were segmented based on the difference of R,G and B values of each target.Compared the results of image segmentation obtained by R,G and B distribution with the results obtained by the multiple threshold and K-means clustering analysis.The results indicated that the image segmentation method based on the R,G and B distribution was better than multiple threshold and K-means clustering analysis.Meanwhile,computing the identification rate of apple on the basis of image segmentation using the method of R,G and B distribution,the results showed that the identification rate was 93.4%.Consequently,it could be concluded that the image segmentation with the help of R,G and B distribution was reliable.