Cotton Plant Dynamic Recognition System Based On The Binocular Stereo Vision Technology

With the full mechanization of the national cotton production, the increasing level of intelligence andthe researching of cotton top that to identify and locate the precise detection technology, Finally, achievethe job of mechanical cutting cotton Top, What is the main steps of cotton mechanization and intelligent ofthe whole. Based on binocular stereo vis ion technology and combine with Xinjiang’s cotton planting statusquo, using the design theory of mechanical systems, pattern recognition, digital image processing, etc,inaugurated a system of cotton dynamic recognition which based on binocular stereo vision technology.The main contents and conclusions are as follows:1. Using Zhang Zhengyou calibration method, combining with OpenCV to compile and develop ofthree-dimensional calibration toolbox and complete the three-dimensional camera calibration, obtainedcamera`s internal and external parameters.2. Using Adaboost algorithm design multi cascade classifier to achieve the cotton top detection andidentification, the rate of detection is83.41%, false detection is8.81%, speed detection up to3fps/s,basically achieve real-time requirements.3. Using digital image processing technology, SIFT matching method, combining with the cameraparameters and the least squares method to calculate the depth information: When the actual distance in therange of1200-1500mm, the error range is3.662-13.341mm, average value is8.795mm, basically achievethe required accuracy.4. Using Microsoft Visual Studio2010development tools, MFC (Microsoft Foundation Classes)library and OpenCV (Open Source Computer Vision Library) computer vision library implementsalgorithms completed stereo vision system of each module, develop and design man-machine interfacevisual system on three-dimensional dynamic recognition of cotton plant.Aiming at the problems which existing in the process of cotton topping by machine, a new method forcotton plant identification based on machine vision technology was post in this paper. The cotton plantidentification system was build, the cotton identification, location, were achieved, which laid thefoundation for subsequent further research.