Classification And Recognition Of Wheat Grains Based On Morphological And Color Features

Wheat is one of the most important food crops in China.Maintaining high yield of wheat has important significance for promoting the rapid development of the domestic economy,improving the people's living standards and the stability of the society.Research on wheat image recognition technology has become a hotspot in the field of agricultural informatization.In this paper,a method for classification of different types of wheat grains based on the morphological and color features is introduced,which can not only save a lot of time,manpower and material resources,but also make up for the shortcomings of manual detection,such as low efficiency,poor objectivity and low accuracy.In this paper,we select five types of wheat grains(Zhoumai 18,Yumai 3,Yunong202,Yujiao 6 and Luomai 29)as the research objects.Firstly,images of wheat grains were captured by a HD digital camera.Then,the images were pre-processed,such as grayscale,enhancement,image segmentation and morphological processing.After that,four shape features(perimeter,area,circularity and elongation)and six color features(R,G,B,H,S,I)of wheat grains were extracted from pre-processed images.Finally,5types of wheat grains were classified by BP neural network.In the study of classification,using only color feature as the input of network.In this paper,there kinds of BP network inputs were tested(using only color features,using only shape features,or using both color and shape features).Experimental results showed that the classification accuracy of the three types of wheat grains,four kinds of wheat grains and five kinds of wheat grains were 95.6%,93.3% and 89.3%.When we used both color features and shape features as the network input,the results were significantly better than using only the color or shape feature as the input of network.The research result showed that the proposed wheat grains classification method is simple to achieve,effective and accurate.Finally,the application of image processing technology in the classification and recognition of agricultural products is also analyzed and prospected.