基于形态学的棉花种子破损检测

Identification and Detection for Surface Damages of Cottonseed Based on Morphology

以"中棉35"棉花种子为对象,应用图像处理技术,提出了基于形态学的破损识别方法。采用链码算子获取轮廓信息并进行傅里叶算子平滑,根据曲率特征定位轮廓尖端点后,通过轮廓的对称性识别局部破损棉种;基于统计思想,根据大小、形状特征的差异识别严重破损棉种。选取完好、破损棉种各200粒进行识别试验,结果表明破损棉种识别率达到94%。

Taking the seeds of ＂Zhongmian 35＂ as research objectives, a method of surface damages detection was developed based on morphology of cottonseed by the digital image technology. The boundary was picked up with chain code operator and smoothed by the Fourier operator. Find the tip of the cottonseed after analyzing the curvatures of contour points. Detect the local damaged cottonseeds according to the symmetry of the boundary. Area and elongation were used to identify seriously damaged cottonseeds with statistical method. The experiments show that the detecting accuracy is 94 % with 200 damaged cottonseeds and 200 cottonseeds without damages.