基于数据融合的玉米种子内部机械裂纹检测方法

Detection of internal mechanical cracks in corn seeds based on data fusion technology

为深入研究玉米种子脱粒和输送等环节中产生的内部裂纹机理和检测技术,该文在体视显微检测基础上提出了基于融合技术的边缘检测方法。该方法采用改进的数学形态学方法和传统Sobel边缘检测算子对损伤玉米种子图像进行边缘检测,建立相应的融合规则,将2种方法检测出来的图像边缘进行基于小波变换的融合处理,并从新图像中提取玉米种子内部机械损伤的特征信息。结果表明,该检测方法结合了2种边缘检测方法的优点,有效提高了边缘检测准确性,在准确提取玉米种子内部裂纹特征同时能有效降低噪声,较单一边缘检测算法有更好的效果。

In order to further study the mechanism and detection technology of internal cracks during threshing process and transportation of corn seeds, an edge detection method with data fusion based on stereomicroscope was proposed. The image edges of corn seeds with mechanical damage were respectively detected by mathematical morphology and Sobel, and fusion rules were set up accordingly. The two results from above methods were then processed by fusion based on wavelet transform to generation a new image. The feature information of inner mechanical damage from new image of corn seeds was extracted. Results showed that the proposed method had the advantages of two edge detection methods, which could improve the accuracy of edge detection and reduce noises while accurately extracting internal mechanical traits of corn seeds. The new method could obtain better effect than single traditional edge detection method.