Rule based concave curvature segmentation for touching rice grains in binary digital images

This paper presents a novel image processing algorithm that has been developed and tested for the accurate segmentation of contacting rice grains in digital images. This algorithm uses general shape properties for initial decision making, then examines concave curvature in the shape boundary to select `split points´ to draw a segmenting `split line´ between contacting grains at the point of contact. This approach offers a cautious, decisive and reliable segmentation between small sets of contacting rice grains. The algorithm has been applied to test images with success in all cases. Results for contacting rice grains are compared with shape descriptors for non-touching grains. It is found that the impact of segmentation on the shape of target grains is negligible. This algorithm is of benefit for intelligent grain analysis.