论文部分内容阅读
为快速、准确地分割黄瓜叶部病害图像,提出一种基于混合颜色空间的双次Otsu算法。算法根据病害图像各部分的颜色特征,首先选取原始彩色图像的R分量进行初始Otsu分割和形态学相关操作,将R分量图分割为背景类和非背景类;然后选取非背景类图像的Cr分量进行第2次Otsu操作,将非背景区域分割为正常叶子类和病斑区域类,得到最终的分割结果。将该算法应用于黄瓜靶斑病图像的分割中,并与R_Otsu算法、H_Otsu算法以及图切割算法进行比较。试验结果表明:与对比算法相比,本算法在分割精度及处理速度2方面的综合分割性能最优,错分率均值和方差分别为2.12%和0.08%,平均处理时间<0.2s,算法对光照变化具有一定的鲁棒性。本研究算法可为自然光照条件下黄瓜病害图像实时、准确分割提供技术参考。
In order to segment the diseased images of cucumber leaf quickly and accurately, a double Otsu algorithm based on mixed color space was proposed. According to the color features of each part of the disease image, the algorithm firstly selects the R component of the original color image for the initial Otsu segmentation and morphology-related operations, and divides the R component into background and non-background classes. Then, the Cr component The second Otsu operation, the non-background area is divided into normal leaves and lesion area class, the final segmentation results. The algorithm was applied to the segmentation of cucumber target leaf spot images and compared with R_Otsu algorithm, H_Otsu algorithm and graph cut algorithm. The experimental results show that compared with the comparison algorithm, the proposed algorithm has the best overall segmentation performance in terms of segmentation accuracy and processing speed. The average misclassification rate and variance are 2.12% and 0.08% respectively. The average processing time is less than 0.2s. Light changes have some robustness. The proposed algorithm can provide technical reference for real-time and accurate segmentation of cucumber disease images under natural light conditions.