Research On Egg Freshness Of Real-Time Non-Destructive Testing Based On DSP Image Processing

In this paper,take the TMS320DM642 digital signal processor of TI as the core, On hardware,the full advantage of the DM642 high-speed data processing capabilities and real-time processing features.By software,In CCS development environment.use the C language to design image processing and feature extraction algorithms.after then,to analysis the egg image under the transmitted light effectively.Separating the characteristics of egg yolk and air chamber.Extracting the ratio of yolk' area and the whole egg's area,the ratio of air chamber'area and the whole egg' area,the ratio of air chamber'height and the egg's long axis,the ratio of air chamber'diameter and the egg's short axis.Take the four parameters as the Non-destructive testing parameters one by one.To find a single optimal characteristic parameters to identify freshness.All processing such as image acquisition,image processing algorithms,image display,program optimization,debugging and recognition process are implemented on the DSP.Detection system with real-time function.(1) DSP hardware platform for learning and application. Combination of CCS software, to study various image processing algorithms.Via video capture, encoding, processing, decoding and display a series of processes, develope an non-destructive egg freshness of real-time detection system.(2) Image processing methods to obtain egg yolk Features:by acquisiting the egg's G component, Grayscale equalization, median filtering, anti-color, corrosion, "and" operation,automatic thresholding and a series of image processing methods to make egg yolk appear.and then,it can gain the ratio of yolk' area and the whole egg's area,that used for non-destructive testing.(3) Image processing methods to obtain air chamber Features:from egg's I component image can be more obviously display the air chamber area.Through the median filter,background replacement,histogram displayed,by the large number of experimental studies to be the optimal threshold segmentation points for thresholding. after morphological denoising, access to regional patterns,according to two regional patterns,dividing into two kind's of algorithms to obtain air chamber partition line basis points to split.This method can cut exactly and speedly.After obtained the air chamber,to extract the ratio of air chamber's area,height,diameter. But this algorithms is restricted in the circulation of actual detection.,the air chamber evenly distributed in the egg big side.(4) Building DSP hardware platform,and Image capture device for testing.Select 30 eggs, At the same time each day over the use of software algorithms to extract the characteristic parameters of four groups of data, Seeking the same day the average of each parameter as a parameter indicator. the other hand,take three eggs,which in the same environment and of the same batch,measured haugh value.obtain the average value as the day's eggs'haugh. until the egg appears yolk rupture.Done a total of 29 days to do tests to track the freshness of eggs.(5) Take the use of SPSS software to establish the four characteristics of value and freshness(Hough value) relational model.Model has a very high correlation coefficient. Upon examination,through the four feature model,the accuracy of the egg s'freshness grades recognition were 91%.92%,92%,91%.(6) With the egg yolk area ratio,air chamber area ratio,air chamber height ratio,air chamber diameter ratio accessed from tests,take them as the input training set samples one by one.other hand,take the corresponding haugh grade as the output samples,building the BP neural network.At last,use the set of validation data to test the reliability and classification accuracy of the network,,the accuracy of the egg s'freshness grades recognition were 90%,89%,90%,89%.(7)Designed an egg freshness interface of real-time detection system based on Visual C++6.0. The results by DSP processing are showed on PC, The PC to communicate directly with the DSP by the system. After starting the DSP,It can get the data by DSP real-timely; also the eggs'haugh value and freshness grades.When the eggs in the transport and storage,Eggs placed not straight,the air chamber shape may be offset,Then the image of air chamber also do not rule.But algorithm in this paper,assuming the air chamber is distributed evenly. And no position restrictions on accessing yolk area ratio.Therefore,a combination of factors to consider, Finally,Chosed yolk area ratio for dynamic images non-destructive testing parameters to identify the eggs'freshness grades.this method can be used online inspection.Through the analysis of test results can be drawn:In this paper, the parameters of non-destructive testing of the eggs has practical reference value, and the system was evaluated by real-time results show that the system in the processing speed and accuracy can basically meet the requirements of real-time detection.