Design Of Cotton Impurity Recognition System Based On Image Processing

China is a big cotton producer in the world.In 2017,China's cotton output reached 5.49 million tons,of which Xinjiang's cotton production accounted for 74.4%.With the continuous development of agricultural modernization,cotton automatic planting and mechanical cotton picking have become the main development trend.Compared with artificial cotton picking,the machine-made cotton yield is obviously improved,but the cotton has a high impurity content.Aiming at the high impurity rate of machine-made cotton in China and the imperfection of impurity detection process,it is of great significance to image the foreign impurities mixed in cotton in machine cotton,to monitor the impurity content in real time and to realize the detection of fast cotton impurities.The impurity detection and analysis technology of cotton impurity image is mainly based on machine vision.According to the different color of the background color of the cotton obtained by the camera,the threshold is divided according to the different characteristics of cotton and impurities in HSV color space,and the impurity characteristics and BP neural network training are extracted.Finally,Calculate the impurity content and compare and analyze the impurity detection system.The program design of cotton impurity detection and analysis system is written by using MATLAB software.Finally,the motor rotation speed is controlled according to the cotton impurity rate,which facilitates the orderly subsequent cotton impurity cleaning process.(1)The construction of machine vision experiment platform is one of the research focuses.The motor drive module and camera module are two important components of the experimental platform.The STM32 and MATLAB software use the RS-232 serial port standard to perform data transmission and Accept,connect the stepper motor driver and camera module through the STM32 development board of the lower position machine,control the motor rotation and cotton image acquisition and transmission,display and save the cotton image in real time in the upper computer to facilitate the next step of cotton impurity image processing.(2)For the impurity characteristics of cotton,select the background plate that does not reflect light,and use the machine vision module to take photos to obtain experimental samples.Compare the HSV color space three-channel histogram and the judgment of the impurity color to determine the best yellow background plate.Image enhancement(salt-salt noise and median filtering)processing is performed on the sample image.According to the HSV color space S component histogram,the difference between cotton and impurity background is analyzed.The background color threshold of the sample is first determined,and the threshold segmentation is performed to obtain the connectivity of the cotton-containing image.The region,extracted impurity characteristics and BP neural network training,and finally calculated the impurity content.(4)Program design of machine vision experiment platform and write CIIAS cotton impurity image detection and analysis system software to realize the extraction of impurity characteristics of cotton impurity image,calculate the impurity content,conduct cotton impurity detection and comparison analysis experiment and control motor speed.The experimental results show that it is best to select the yellow background plate by RGB color space,and distinguish the different color characteristics of cotton and background according to the S component saturation of HSV color space,which is beneficial to extract the impurity contour features,HSV threshold segmentation and impurity ratio average.The error is 3.39%,and the average error of the BP neural network to identify the impurity content is 2.60%.The image processing has a high efficiency and accuracy of 97.4%.The CIIAS cotton impurity image detection and analysis system software basically fulfills the task requirements of impurity detection and analysis.