Automatic segmentation of touching rice kernels with an active contour model

An approach was developed to segment touching rice kernels in an image. An inverse gradient vector flow (IGVF) was first proposed to automatically generate a field center for individual rice kernel in an image. These centers were employed as the references for setting initial deformable contours that were required for building an active contour model (ACM). In addition, to describe the formation of the initial deformable contours in detail, a complete image process for the segmentation of touching rice kernels was also developed. The result showed that as long as 50% of piecewise edge information remained in an image, the algorithm could reconstruct the whole contour successfully. Compared with the original contours, the contours generated in this study achieved more than 96% similarity. The complete contours of touching objects by the approach proposed in this study could facilitate the subsequent image processing to obtain the geometric, texture, and color characteristics of objects in an image. These features might then be used for further clustering, classification, or image understanding.