Research And Test Of Moldy Peanut Kernels Based On Color Sorting Technology

Peanuts are one of the most important oil crops in the world and second only to rape planting area,which occupies an important position in world food markets.Peanuts are rich in protein,fat,various comprehensive and relatively balanced nutrients,which is the ideal high protein,high fat nutritional food source.Peanuts have been inevitably deformity,impurities and mildew leading to lower the quality of the peanut and commodity value during the process of growth,harvest,storage.Early separation peanut mostly rely on artificial sorting which are issues such as high production costs,low production efficiency,low accuracy and easily influenced by subjective factors.However,there is no mature peanut specifically for localization of photoelectric color optional equipment in our country.At the same time,the imported equipment is too expensive.Therefore,peanut kernel color sorting experimental study has urgent demand and important practical significance.Main contents and results:(?)With the research object of color sorter,the operating parameters of quantity,blowing time,sensitivity,the objective function of color sorting accuracy and ratio of color separation mass,the paper builds the multivariate regression model analysis explore the influence of various factors and the best level combination.Use Design-Expert8.0.6 software regression analysis and response surface analysis model to optimize the optimal operating parameters peanut color sorter.The best combination of operating parameters is quantity of 36,blowing time of 3ms,sensitivity of 76,corresponding to the color selection accuracy of 97.54%,with a ratio of 14.66.And every performance optimization theory and the relative error is less than 2%.The results can provide a reference for optimizing the operating parameters peanut color sorter.(?)This paper puts forward a sorting algorithm on the basis of image processing technology according to the characteristics of peanut kernels image,and carries O experimental verification for this algorithm through MATLAB software.Peanut kernel image preprocessing model consists of the following five steps.First,the segmented stretch direct gray level transformation method is utilized to carry O image intensification for gray level image.Second,the median filter is utilized for image filtering.Third,the iterative method is used for image segmentation.Fourth,carry out the morphological algorithm of closing operation.Fifth,the image synthesis is completed by pixs point correspondence converter method.Obtain the color peanut kernel image after image preprocessing.(?)Extract the distinctive color characteristics of peanut kernel in the image (?),(?),(?),which are taken as input characteristic parameters of BP neural network.Code(0 1)and(1 0)to represent the network Oput results of normal peanut and moldy peanut,and establish manual neural network sorting model.According to the experimental results,the average sorting accuracy is 98.82%.T test verifies that there is no significant difference between moldy peanut kernels results of the manual sorting and image processing sorting.A favorable sorting result is obtained,which has the certain reference value for the real time moldy peanut kernel sorting equipment research.(?)The advantages and disadvantages of the current mainstream embedded chip are analyzed.The Texas instruments(TI)specifically for C6000 series DSP TMS320C6748 chip development in image processing are selected as a system of master control chip in the end.And the paper makes a brief introduction of the on the TMS320C6748.The main image processing algorithm function contains RGB to gray image transform,the RGB to HIS image transform,the segmented stretch direct gray level transformation method,the median filter,the iterative method model forimage segmentation,and extract the characteristic parameters,and initialize the BP neural network,BP neural network training,the Oput of BP neural network,the feedback learning of the BP neural network.The C compilers of TMS320C6748 support the ANSI C standard C language program design.In order to achieve this research moldy peanut kernel image processing algorithm is realized in DSP embedded system,the main control chip embedded system are selected,moldy peanut kernel image recognition algorithm implemented by C language and the specific code are analyzed.