基于混合蛙跳算法的马铃薯病害图像分割优化

针对Otsu算法对直方图呈现多峰多谷的复杂马铃薯病害图像分割效果不佳的问题,结合混合蛙跳算法(shuffled frog leaping algorithm,SFLA)提出了一种Otsu-SFLA分割优化模型,将Otsu病害图像的分割结果作为SFLA算法的优化起点,进行复杂背景的马铃薯病害图像的分割优化,将马铃薯叶枯病、马铃薯晚疫病、马铃薯菌核病、马铃薯根腐线虫病、马铃薯灰霉病的病害图像作为分割对象进行分割,分割匹配率分别为97.0%、96.2%、96.9%、95.7%、94.8%,平均分割匹配率为96.1%,错误率分别为1.6%、1.1%、1.2%、1.1%、1.4%、平均错误率为1.3%,正确率分别为95.4%、95.1%、95.7%、94.6%、93.4%,平均正确率为94.8%,表明Otsu-SFLA模型可有效从复杂马铃薯病害图像中获取病斑区域。

Segmentation optimization of potato disease images base-on shuffled frog leaping algorithm

According to the problem that the Otsu algorithm does not have a good segmentation effect on the complex potato disease images whose histogram presents multiple peaks and valleys, an Otsushuffled frog leaping algorithm (Otsu-SFLA) segmentation optimization model was put forward in this paper, assuming the segmentation result of Otsu algorithm to be the original value of SFLA optimization to start the next optimization. In order to verify the validity of this model, leaf blight, late blight, sclerotium, root rot nematode and gray mold of potatoes were taken as the test segmentation targets, the matching ratio of segmentation could reach 97.0%, 96.2%, 96.9%, 95.7%, 94.8%, respectively, the average matching ratio was 96.1%, the error ratio could reach 1.6%, 1.1%, 1.2%, 1.1%, 1.4%, respectively, the average error ratio was 1.3%, and the correct ratio could reach 95.4%, 95.1%, 95.7%, 94.6% and 93.4%, respectively, the average correct ratio was 94.8%. The test result demonstrated that Otsu-SFLA model can effectively extract the disease symptom area from complex potato disease images.