| With the continuous maturity of computer technology,the automatic processing based on image processing technology has become more widely used in different research fields.However,at present,the appearance quality inspection of rice in China is still at the stage of artificial naked eyes inspection,which is not efficient,and at the same time,the test results are subject to errors due to excessive subjectivity.This not only causes great damage to consumers’ health,but also results in China.Rice’s competitiveness has decreased.Therefore,the automation of rice appearance quality testing has become an inevitable trend of social development,which is of great significance for national development and social food security maintenance.Because of the adhesion between the collected data samples,at the same time,for the manual detection,the detection results of chalky and yellow grain rice have errors.Therefore,this article will focus on the three issues of sticky rice segmentation,chalky detection and yellow grain detection.In the segmentation of sticky rice,a Gaussian mixture model segmentation method based on density peak clustering is proposed.This method first calculates the local density and distance values through the density peak clustering algorithm,and uses the improved ranking map to replace the original decision The map then determines the potential cluster centers,and then substitutes the potential cluster centers as the initial cluster centers of the Gaussian mixture model for calculation,and calculates the corresponding contour coefficient value.Finally,the value corresponding to the best contour coefficient value is Gaussian mixture model results are used as the final segmentation results.The experimental analysis shows that the algorithm has a more effective segmentation effect for rice adhesion than the traditional watershed algorithm,pit detection algorithm and K-means algorithm.In solving the problem of chalkiness detection,this paper proposes a spatially cooperative chalkiness detection method based on color quantization,which fully considers the color information and spatial structure characteristics of chalkiness.The algorithm first quantizes the color of each grain of rice obtained by segmentation,then calculates the color similarity between the "chalky" part and other color parts,and performs ellipse fitting on the part to calculate two reflecting the spatial structure of the part Relevant parameters,finally,chalky detection is performed according to the above two conditions.The experimental analysis proves the feasibility of the algorithm in the chalky detection process.In terms of yellow rice detection,this paper has made some improvements to the detection method based on HSI color space,combining the concept of color quantization.By comparing the quantized color histograms,it is found that the difference between normal rice and yellow grain rice is more obvious.The experiment proves that the improved yellow grain rice detection method in this paper can more effectively achieve yellow grain rice detection. |
