The Study On The Segmentation Of Grain Particles Images

Nowadays, the world grain problem is one of the urgent problems that attract worldwide attention. To guarantee the security of Chinese grain, we should pay more attention to the improvement of grain production, grain storage and Circulation System. Grain quality analysis is one of the key technologies to guarantee food security in our country. By digital image processing technology can avoid some defects of low speed, low accuracy and poor repeatability which exist in the grain appearance detection correspond to manual method. Image segmentation technology have become one of the hottest and most difficult subjects of image processing technology study, quality of image segmentation directly impacts on the follow-up operation of the image. This paper mainly studies the issue of grain image segmentation.For the particles adhesion image segmentation issues, the paper first put forward an improved method of mathematical morphology, which is basis on the doctoral dissertation of LinYun. The experiment showed that this method can effectively separate adhered particles, the effect will be even more obvious when the contact area of particles are smaller and the broken rice ratio are higher. Lastly, in view of traditional watershed segmentation method tends to appear over-segmentation phenomenon when partitioning particles adhesion images, a modified method of watershed segmentation that the button threshold controllable is proposed in the paper. It's lead in the method of edge-detection. We draw a conclusion that the algorithm can serves as better and faster solution to segment various adhesion level of grain particles image, and eliminates the over-segmentation through a large number of experiments. In the end, this paper analyzes the broken rice ratio using digital image processing technology. These methods are validated in improving the speed and precision of broken rice ratio detection by experiment.