摘要
采用Sobel等4种常用方法检测长翅灰飞虱翅膀、爪子等多细节边缘时,会出现边缘不明显、噪声干扰大等问题。采用形态学单结构腐蚀边缘检测也不理想,其边缘间断数和像素损失率最小值分别为10和0.689%。为此,对长翅灰飞虱图像先进行区域划分,以使单个区域的检测得到简化,并针对划分后的各个区域采用形态学多结构方法构造了膨胀腐蚀型边缘检测算子进行边缘检测。试验结果表明:这种划分区域并结合多结构的方法能提高对长翅灰飞虱边缘的检测能力,其边缘间断数和像素损失率最小值分别为3和0.554%。
When the four conventional edge detection methods such as the Sobel are applied to detect the multiple-structure edges of the Long Wing Laodelphax Striatellus （Fallen） image including the wing and the claw, it will be influenced by the pixels noise and the edge will not be clear. Also the morphology of erode single element edge detection method are adopted and the result is not ideal, as the breakpoints is 10 and the minimum loss rate of the pixels is 0.689%. In order to improve the ability for detecting the edge, the image was divided into several areas due to the multi-edges so that the edge detection in every area was simplified. Subsequently, the erode with dilate edge detecting operators was constructed, which could be used to detect the different edges in every areas. The results indicate that the multi-areas combined with morphology multi-structures method for detecting the Long Wing Laodelphax Striatellus （Fallen） edge is better, and the breakpoints is 3 and the minimum loss rate of the pixels is 0.554%.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2008年第7期119-123,共5页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金资助项目(项目编号:30571240)
江苏大学高级人才启动基金项目(项目编号:05JDG040)
关键词
长翅灰飞虱
边缘检测
多区域
多结构
Long Wing Laodelphax Striatellus （Fallen）, Edge detection, Multiple areas,Multiple structures
