论文部分内容阅读
针对脐橙自动分级检测中存在正确识别率偏低、实时性不强的问题,提出一种综合特征提取方法:在对图像颜色模型进行转换后,用H分量图像提取脐橙的大小特征;S分量图像通过背景分割、边缘灰度补偿、整体亮度变换后提取脐橙的果面缺陷特征;采用R、G、R-G3个分量的均值和标准差提取脐橙的颜色特征。以脐橙的大小特征、果面缺陷特征和颜色特征为支持向量机(Support vector machine,SVM)的试验输入向量,进行脐橙分级检测试验,以实现提高脐橙自动分级正确识别率和增强实时。试验结果表明:该SVM分类器对脐橙分级的正确识别率为91.5%,处理时间为160ms,适合于实时环境下的分级检测。
In order to solve the problem of low accuracy and poor real-time performance in automatic grading detection of navel orange, an integrated feature extraction method is proposed. After the image color model is converted, the size features of navel orange are extracted by using H component image. The fruit defects of navel orange were extracted by background segmentation, edge gray-scale compensation and overall brightness transformation. The color features of navel orange were extracted by using the mean and standard deviation of the three components of R, G and R-G. The navel orange size test, the navel orange size test, the fruit defect and the color feature were used as input vectors of Support Vector Machine (SVM) in order to improve the automatic grading accuracy and real-time performance of navel orange. The experimental results show that the SVM classifier has a correct recognition rate of 91.5% and a processing time of 160ms, which is suitable for grading detection in real-time environment.