Maize seed embryo and position inspection based on image processing

Abstract

Orientational planting cannot only make maize leaf growing consistently, but also improve the leaf photosynthetic capacity of unit area and yield of maize. The "tip" direction was detected and the angle of deflection was measured by using contour curvature analysis method of maize seed, the S component of HSV channel was preprocessed by Otsu method, the automatic extracting maize seeds embryo and orientation information have been realized through image color channel conversion, segmentation, preprocessing, and its contour characteristic analysis. A rectangular region ROI in the image was defined and counts up pixels within the region, the region-specific positive and negative ROI pixels were compared by TM threshold and the embryo side towards was identified. This paper adopted Zheng-958, Jundan-20 and Zhongke-11 maize seed for research object, each variety was repeated three times by using the above methods. The results showed that the average accuracy of embryo inspection was more than 95%; the direction average angle was 2.2 °.