Research On Detection Method Of Raw Cotton Impurity Based On Image Processing

The presence of impurities in raw cotton will seriously affect the appearance and quality of cotton products.Therefore,it is very important to detect and identify impurities contained in raw cotton,which is the premise of removing impurities.Impurities in raw cotton have less content,but they have many types and are heterogeneous,so it is difficult to detect them.At present,the research of automatic detection through the combination of image processing and machine vision has become the focus of current research.This paper takes the raw cotton impurity image as the research object,and proposes an algorithm for raw cotton impurity detection through the design and experiments of related algorithms.Firstly,in order to reduce the loss of information of the original image during the grayscale conversion process and retain more image details for subsequent processing,based on the global mapping algorithm,an adaptive allocation of R and G based on the attributes of the image is proposed.Compared with the traditional weighted average algorithm,the grayscale algorithm with the three component weight ratios of B and B can retain an average of 5.5% of the image information after the grayscale processing;and then perform homomorphic filtering to adjust uneven lighting and segment The linear transform enhancement processing and median filter denoising processing are used to improve the image quality and reduce the adverse effects of external factors on the image.Then,in view of the complicated and tedious threshold selection process of the traditional Canny algorithm,the threshold selection part is improved locally by introducing the idea of the largest inter-class variance.The improved algorithm can not only adaptively generate relevant thresholds based on different image characteristics At the same time,the high and low thresholds are no longer a simple and fixed multiple relationship,and the improved algorithm has a better segmentation effect.In the subject of image recognition,five types of common raw cotton impurities,including hair silk,feathers,plastic ropes,shell leaves,and small neps,were selected for feature value extraction and recognition.It mainly researches the related color characteristics and shape characteristics of five types of impurities.Through analysis and research on different characteristic parameters,two color characteristic value parameters are finally selected among many characteristic parameters: the L component in the Lab color model.Value,B component value in the RGB color model;2 shape characteristic parameters: eccentricity,area perimeter ratio,comprehensively determine the selected five types of impurity types through the distribution interval of 4 characteristic parameters,and finally complete the detection of impurities And identification.The designed algorithm is finally tested,and the obtained results are compared with the actual results.It is found that the accuracy of the algorithm designed in this subject can reach 91%,especially the recognition effect of hair silk and plastic rope is the best.