基于支持向量机的缺陷红枣机器视觉识别

Recognition of Defect Chinese Dates by Machine Vision and Support Vector Machine

在枣的干制过程中形成的油头枣、浆头枣、霉烂枣是最常见的缺陷枣,它们整体或局部颜色偏暗、偏黑,有必要通过机器视觉技术将其识别出来。在HIS颜色空间中,提取H的均值和均方差作为红枣的颜色特征值,利用支持向量机识别缺陷红枣。实验结果表明,识别准确率可以达到96.2%,优于人工神经网络的89.4%。

During the production and storage of Chinese dates, some of them are easy to mould rot because of high water content. The defect dates appear darker than the normal ones. Based on support vector machine, the recognition of the defect Chinese date machine vision was proposed. After the acquisition of the Chinese dates images, the color model was changed from RGB to HIS. Then, the average value H and standard square deviation value aH of dates hue values were calculate. Depending on the two values, there was few overlaps between defect dates and normal ones in the plot of H and all. Therefore, H and aH were treated as the feature parameters. Artificial neural network （ANN） and support vector machine （SVM） model were used to analysis the dates features respectively. The experimental results show that SVM has a better performance than ANN on distinguish defect Chinese dates from normal ones, and the correct recognition rate of SVM is 96.2 %.