Research On The Algorithm Of Cotton Image Segmentation And Maturity Judgment

Xinjiang is a province which has high cotton production and acreage. It’s cotton acreageoccupy more than one-third of the country’s total area. However, a large part of the Xinjiangcotton picking rely on manual labor, it not only restrict the cotton picker efficiency and labor costs,but also have some impact on the social environment in Xinjiang. The emergence ofcotton-harvesting robot can not only improve the efficiency, saving labor resources, but alsoovercome the shortcomings of mechanical cotton picker high quality. Therefore, the research anddevelopment of the cotton-harvesting robot will be the future development direction, it have broadapplication prospects. The cotton image segmentation and discriminate of maturity is a keytechnology for cotton-harvesting robot’s visual technology research, this article focuses on thistwo aspects.The light intensity of Xinjiang is very high and cotton image have uneven brightness. Itcaused great difficulties in segmentation. To this end, this paper presents a new algorithm ofcotton image segmentation. The methods extract the sample pixel value from the two categoriesunder OHTA color space. At first we use Otsu method to deal with the image. In the following, weuse the support vector machine to establish classifiers, and use it to segmented image. In the end,we removing the noise by region label. The experiment result shows that this method can segmentcotton image with the complex background, and the accuracy of segmenting is as high as92.3%.What’s more, it’s classify speed is better than directly using SVM to segment the image. Theaccuracy and stability are better than threshold segmentation method.Before picking decisions, we need to determine the maturity of cotton. This paper proposedan algorithm based on the shape feature. The method is extracted the shape feature from binaryimage. We select five characteristic parameters through statistical analysis, such as aspect ratio.Then use SVM to develop the discriminate model. The optimal parameters of RBF kernel functionwere optimized by the algorithm of PSO. Finally, we use model to determine the maturity ofcotton. This The experiment result indicates this method can determine the maturity of the cottoneffectively in a high-speed, it has a great practical value. The maturity of the cotton effectively in ahigh-speed, it has a great practical value.