The Research For Color Sorting Technology Based On Color Linear CCD

With the development of computer vision, intelligent control technology,industry intelligence has been fully used during the processing of agriculturalproducts. Rice color sorting technology and rice color sorter development provides aplatform for people in the pursuit of high quality rice. The image informationcollection system and the sorting algorithms have been designed and studied in therice sorting system in this project.The color image sensor and sorter technology research at home and abroad ofrice color sorter has been introduced in this paper. And then design the rice imageacquisition system based on color linear array CCD. FPGA is used as a master designpart, mainly CCD drivers, AD-driven and designed the cache module FIFO, ADmodule will convert processed information to FIFO, and designed the USBcommunication module, by FPGA Module control USB connection module with thePC machine. Designed the image denoising and sorting method based on PCmachine.This common space for color images were analyzed, which includes RGB space,CMY space, CIE space and YIQ space, as well as HSI space, through analyzingchose HSI space as the selection design space in this paper. Because of the lineararray CCD color image information collected includes noise problem, this paper alsoanalyzes common noise in an image which is the Gaussian noise, salt&pepper noiseand mixed noise. And study the denoising methods based on the characteristics ofnoises, and through PSNR to judge the effect, select a suitable method to completeimage denoising.Through the study on fuzzy mathematics and cluster analysis in this paper, andthen study the fuzzy c means clustering algorithm, the rice images sorting algorithmcombined with the HSI color space has been designed. Through the algorithm simulated in the MATLAB environment, finally proved that the algorithm can bewell separated the normal rice and different color rice between the rice images.Especially to distinguish the different color point in a rice image has a higheraccuracy.