遗传算法改进的KSW熵法计算黄瓜叶部角斑病密度

徐海 秦立峰

摘要[目的]去除复杂背景影响，提高角斑病病斑分割精度和速度。[方法]首先对预处理后的b*通道图像采用大津法进行初分割，去除大部分背景和噪声。再对目标部位的灰度图，用基于遗传算法改进的KSW熵阈值分割法进行二次分割，得到病斑的二值图像，并计算病斑面积，最后与叶片面积做比得到病斑密度。[结果]该方法计算的病斑密度与方格板手动计算的结果的绝对误差约为0.02，而病斑的分割速度提高了45%以上。[结论]该方法为黄瓜角斑病病害程度自动诊断提供技术依据。

关键词黄瓜角斑病；图像分割；遗传算法；KSW熵

中图分类号S126；TP391.4文献标识码

A文章编号0517-6611（2017）32-0212-04

KSW Entropy Method Improved by Genetic Algorithm for Density of Cucumber Angular Leaf Spot Calculation

XU Hai1，QIN Lifeng1，2*（1.Electromechanical College，Northwest A& F University ，Yangling，Shaanxi 712100； 2.Key Laboratory of Agricultural EPC System Network， Ministry of Agriculture，Yangling，Shaanxi 712100）

Abstract[Objective] The aim was to eliminate the influence of complex background， and improve the precision and speed of angular leaf spot segmentation.[Method] Firstly， the Otsu method was used to segment the b* channel image after preprocessing to remove background noise.Then， the gray image of the target area was segmented through the GAimproved KSW entropy thresholding method， and the binary image of disease spot was obtained as well as disease spot area was computed.Finally， the disease spot density was calculated.[Results] The results showed that the absolute error of the method calculated by the algorithm was about 0.02 compared with grid plate manual calculation method， and the segmentation speed increased by more than 45%.[Conclusion] The method provides the technical basis for automatic diagnosis of disease degree of cucumber spot disease.

Key wordsAngular leaf spot of cucumber； Image segmentation； Genetic algorithm； KSW Entropy

蔬菜病害對生产者造成重大经济损失，是制约蔬菜种植业发展的重要原因。黄瓜角斑病是黄瓜主要病害之一[1]，其在叶部产生病斑数量和大小是判断病害程度的主要依据。依靠人工观察计数进行判断，费时费力，准确性低。利用图像处理和模式识别技术可实现角斑病病斑密度的自动统计，作为病害防治措施的依据，从而高效地控制病害发生，减少经济损失，是农业信息化技术的重要研究方向[2-5]。

病斑密度计算以病斑分割、叶片与病斑面积计算为前提。葛婧等[6]利用玉米小斑病的RGB图像分两步将病斑分割出来，并利用分割结果求得玉米冠层危害程度。但该研究针对图像背景较单一，不适用于田间复杂背景情况。田有文等[7]研发基于Android平台的黄瓜叶部病害严重程度实时无损检测系统，采用迭代式阈值法分割叶片区域和背景区域；……