Research On Color Recognition In Color Selection For Particle Grain Material

Color recognition is an important foundation and technique of evaluating the quality of products, it is applied to grain production and food processing pervasively. At present, most color sorters used in China are made in abroad because the domestic technology of color recognition is not perfect enough. The paper proposes machine vision as the main technology and particle grain with typical color as the object of research, synthetically applies the knowledge of iconology, optics and computer, takes photos by industrial CCD camera in LED and natural light, and recognizes the color of particle grain at length.The paper mainly researches the methods of color recognition on the basis of the constructed machine vision system. Firstly, images of various particle grains are taken by CCD camera. Secondly, each image is smoothened and filtered by means of median filter in order to make the background noise eliminated, and then sharpened by Laplacian operator. Thirdly, transform RGB color model of particle grain image into HSI color model and then divide it into H, S and I in order to obtain chroma information as color feature. Finally, train the chroma information of several particle grains by means of auto-organizational neural network to obtain the corresponding chroma criterions. And this neural network will distinguish impurities from normal particles according to these chroma criterions. Experimental results show that machine vision is able to recognize the particle grains with abnormal color effectively and precisely. And conclusions of this paper lay the foundations of designing color sorter.