Research On The Pattern Recognition System Of Egg Embryo Based On Support Vector Machine

The pattern recognition system of egg embryo based on Support Vector Machineis studied in this paper. According to the theory of pattern recognition system and theanalysis of pattern recognition systems constitute determine the main content of thepattern recognition system of egg embryo. Build image acquisition system based onmachine vision to acquire egg embryo image, and use the Visual C++software toprogram the image processing program, and design appropriate classification based onalgorithm of support vector machine. Automatic on-line detection of egg embryo canbe achieved through the combination of machine vision system and image detectiontechnology and the combination of machine learning algorithm and pattern recognitiontheory.Based on the machine vision system to design and build the image acquisitionplatform, select the LED light source and the CCD industrial camera to obtain internalimages of the egg embryo, use PLC to control the transmission of egg embryo and themobile of the industrial camera, the selection of Advantech Technology610L typeindustrial control machine is responsible for camera of the camera, image storage andimage processing, and the communication with the PLC while realize human-computerinteraction, completed the automatic acquisition of the egg embryo image.Analyzing and processing of the collected egg embryo, through the using ofsmooth image and image enhancement to eliminate noise in the image, and highlightthe information of the characteristics. Then the amount of information of the vascularand the black block characteristics regional is observed and counted through theinformation display module, and analyzes the histogram information and histograminformation of the generated R, G, B components, through threshold segmentationextract the main characteristics of the blood vessels and the black block in the eggembryo image.According to the design principle of the egg embryo classifier, adopt the algorithm of support vector machine, and select the kernel function of SVM classifier.Based on the characteristic data of the training sample of egg embryo under LIBSVMtoolkit, find the optimal parameters of the classifier in the use of cross-validationmethod, and determine the egg embryo classifier model based on SVM by training.The experimental verification process of the pattern recognition system of eggembryo has been done in combination with the hardware platform, for thecharacteristic data of unknown egg embryo and the performance of the egg embryoclassifier model based on SVM. Calculate the accuracy rate of the SVM classifier, bycontrast the classification results of SVM classifier and the results of manualinspection.