论文部分内容阅读
针对自然光照条件下果园苹果识别效果不佳的问题,从苹果的颜色分割和形状提取2方面进行对比研究,提出一种自然光照条件下的苹果识别方法。利用错检率、漏检率和处理速度3个量化指标综合对比分析颜色阈值、SVM和BPNN 3种苹果颜色分割方法的处理效果。比较6种边缘检测算法对苹果区域图像的边缘检测效果,并使用Hough圆检测算法对苹果形状进行提取,以获得苹果的圆心和半径。试验结果表明:由BPNN的苹果颜色分割方法以及结合Log和Hough的苹果形状提取方法所构建的果实识别算法具有较高的鲁棒性和准确性,能有效克服果实遮挡、重叠和颜色变异等问题,果实平均识别率可达91.6%。
Aiming at the poor recognition effect of orchards under natural lighting conditions, this paper compares the apple’s color segmentation and shape extraction, and proposes a method of apple recognition under natural lighting conditions. The results of color segmentation, color threshold, SVM and BPNN color segmentation were compared and analyzed using three quantitative indexes: mis-check rate, missing check rate and processing speed. Six kinds of edge detection algorithms were compared to detect the edge of the image in the apple region. Hough circle detection algorithm was used to extract the shape of the apple to obtain the center and radius of the apple. The experimental results show that the fruit recognition algorithm constructed by BPNN apple color segmentation method combined with Log and Hough apple shape extraction method has high robustness and accuracy and can effectively overcome the problems of fruit cover, overlap and color variation , The average fruit recognition rate of 91.6%.