Correlation Analysis Between The Appearance Quality And Protein Content Of Soybean

In our country, a variety of crops to plant production, is widely used and has important status more soybeans. The protein content of soybean is whether can represent the soybean nutrition is enough, to make the soybean food index of whether it is good for the human body. Life indispensable cooking oil or soybean milk is based on many aspects, such as soybean protein as raw materials extraction processing, the high quality soy protein can lower cholesterol, not only inhibit blood pressure, reduce blood sugar, anti-cancer, etc., and easy to digest absorb and antioxidant. In order to increase the protein content in soybean to a level, can satisfy the demand of which is to reach the optimal quality breeding ac hievements, some important properties and characterization of soybean can improve together, for soybean every appearance of the correlation between protein content and analysis has become an important research subject. This study application of machine vis ion technology and image processing technology to nondestructive testing of the appearance quality of soybean, avoids the soybean samples of image when the impact of external factors, improve the soybean samples images without interference. Using MATLAB software and SAS software implementation, using multiple linear regression analysis method to the appearance quality traits and protein content of soybean in regression analysis, the regression equation are obtained.Nondestructive testing before the appearance quality of soybean, soybean’s first appearance in the form of characteristic data, the data extraction can exactly the appearance characteristics of soybean is particularly important. Soybean appearance features of this research in addition to conventional shape, color and texture feature extracting, also puts forward a kind of based on wavelet moment characteristics of external quality of soybean extraction method, Of the algorithm proposed is to reduce the noise in the process of feature extraction of soybean, the effects of effectively solved due to the size of the soybean itself, mobile characteristics of unknown, noise pollution and other issues, to provide more accurate sample character correlation analysis characteristics. The concrete research co ntent of this study are as follows:(1) Build a data acquisition system of external quality of soybean. Clear the function of each component and tasks, collect images of soybean; Of image preprocessing, including contrast-enhanced, filter out noise, find edges, divided four steps, preparation for the feature extraction.(2) Application of MATLAB software to extract more than 10 kinds of thousand grain shape characteristics of normal soybean, such as soybean grain area, centroid, perimeter and other characteristics of a total of 12 elements; Color features, such as soybean grain color of RGB color model, such as the mean, standard deviation and three moments features a total of 36 elements; Texture characteristics, such as soybean grain texture energy such as mean, standard deviation and roughness features a total of 12 elements; And this study proposed wavelet moment features, a total of 61 soybean appearance quality characteristic parameters.(3) SAS software, the regression analysis method is applied to the appearance of the soybean quality characteristics data correlation with protein content, protein and circular degree, tightness, such as the correlation coefficient is not zero, is a positive correlation, and standard deviation of R, G, standard test inspection coefficients for the poor, says strong significance; Protein and centroid, perimeter correlation coefficient is not zero, is a negative correlation, and the energy of the mean and standard deviation of moment of inertia test inspection coefficients for, said significantly. Get on in the last R H mean and standard deviation of wavelet moment characteristics, the regression equation between the and the equation is verified.