基于优选傅里叶描述子的粘连条锈病孢子图像分割方法研究

摘要

小麦条锈病越夏孢子数是判断条锈病是否爆发的关键参数之一,现有方法对粘连条锈病孢子的自动检测尚有困难.为了实现粘连条锈病孢子的准确计数,提出一种融合K-means聚类算法与优选傅里叶描述子的粘连条锈病孢子分割算法,并实现了粘连孢子的准确计数.利用K-means聚类算法进行条锈病孢子的分割,实现了包含杂质等复杂背景下孢子的准确分割;对不同傅里叶描述子的边缘描述效果进行评价,优选出适合的傅里叶描述子个数,解决K-means聚类算法所得到的孢子边缘轮廓不够平滑的问题;利用基于距离测度的角点检测算法,实现了粘连条锈病孢子的准确计数;为了验证该方法的有效性,对20幅粘连孢子图像进行了分割实验.实验结果表明,得到的条锈病孢子计数准确率为96.2%,可以用于复杂粘连背景下孢子的准确分割,表明将该方法用于小麦条锈病越夏孢子的计数是有效的、可行的.%The amount of wheat stripe urediospore is one of the key parameters for stripe rust incidence, and the automatic detection of adhesive stripe rust spores is found to be difficult.A segmentation algorithm for adhesive stripe rust spores based on K-means and optimized Fourier descriptors is proposed to realize the accurate counting of adhesive stripe rust spores.With the background containing impurities, the accurate segmentation of stripe rust spores is accomplished by K-means clustering algorithm.The different edge description effect of Fourier descriptors is assessed to optimize the number of appropriate Fourier descriptors, solving the problem that spore edge segmented by K-means clustering algorithm is not smooth enough.By calculating the distance between the smoothed contour points and the given centre point,the corners are detected and used for counting the number of spores.20 adhesive spore images are used to verify the effectiveness of the algorithm, and results indicate that the accuracy rate of spores count obtained by this method is 96.2%, and it can be used for accurate segmentation of adhesive spores with complex backgrounds.The results also demonstrate that the method is effective and feasible for the detection of uredospores of wheat stripe rust.