Research Of Orange Quality Classification Technology Based On Computer Vision

The quality classification of fruits like oranges an important tache between picking and going on sale. It is very important that the research of use computer vision technology to implement automatic detect and classification of orange quality.Suzhou Dongshan orange is the research object in this paper.Based on pretreatment of orange image, emphasis research technical line and implementation methods of use computer vision technology to implement automatic detect and classification of orange quality, include the methods of pick up orange and its characteristic value, classification algorithm.The main achievement of this paper is:(1)According to the characteristics of orange image,based on the analysis of some calculus operator in edge detection, design an orange image edge detection method based on Canny operator. On this basis, combined with human visual characteristics, design color characteristics description methods based on HIS model and shape characteristics description methods based on Fourier descriptor operator. The experiment result proof that Canny operator is high SNR high accuracy and low computation;HIS model is more accord with human vision and low computation also; shape characteristics description methods based on Fourier descriptor operator is more easy to shape classification.(2)Based on the analysis of traditional BP Neural network algorithm that high iteration and complex calculation, design optimization BP Neural network algorithm parameters, use color and shape characteristics values as input data, divide orange four level, The experiment result indicate accuracy rate reach 95%.(3)The author design a method pick up grain characteristics based on Gabor wavelet, combine PCA and SVM to classification of orange.first this method use Gabor wavelet Generating function, and pick up Directions and scales factor to description grain characteristics, then use PCA to reduce dimensions for grain characteristics, formation optimize grain characteristics, On this basis, classification of orange based on SVM algorithm. The experiment result proof that integration BP Neural network algorithm classification result and Gabor wavelet, combine PCA and SVM algorithm classification result, then use simple weights to classification of orange, the accuracy rate is more high reached 97.5%.