| With the further development of modern science and technology,the demand for intelligent equipment in various countries continues to increase.Among them,unmanned boats,as a new member of unmanned systems,have more and more applications in maritime patrols,mineral exploration,surface patrols and other fields.As the technology of unmanned boats matures,the recovery technology for unmanned boats is also continuously improved.Among them,the use of computer vision to realize the recovery and guidance of unmanned boats is a powerful and effective means.At present,the development of computer vision is constantly innovating and progressing.As a member of many intersecting fields,binocular vision technology has received extensive attention from all walks of life since its formation.Among them,binocular measurement is the most widely used.distance.The function of binocular ranging is to find the corresponding feature points from the images of different angles,and then recover the three-dimensional spatial information from the plane images.This thesis takes the UAV recovery guidance as the research objective,and at the same time,considering the impact of haze weather on the image,using binocular cameras,the image defogging algorithm and the ring range measurement in the UAV recovery system are studied.The main tasks completed in the thesis are as follows:The camera calibration and stereo correction are studied.By analyzing and comparing the pros and cons of several calibration methods,the camera was calibrated by Zhang Zhengyou’s calibration method.The image is stereo-corrected according to the calibrated camera parameters,and finally two left and right images that meet the epipolar constraint are obtained,which reduces the complexity of the subsequent stereo matching calculation.Perform preprocessing operations on the image.Histogram equalization for image contrast enhancement,and the median filter is used to eliminate the noise while retaining the detailed information of the image.Aiming at the interference of haze weather on the collected images,the image defogging algorithm is improved.First,convert the original image from RGB to HSV color space domain,perform power exponential calculation on the S and V components that are greatly affected by the haze,and then perform the corresponding stretching,after the calculation,multiple sets of dehazing images are obtained,and then used The cumulative sum of multiple indicators is used as the evaluation standard,and the optimal power index combination is selected to obtain the final dehazing image.The experimental results show that the algorithm effectively removes the haze information in the image.Research the matching cost in the stereo matching step.Aiming at the problem of the traditional Census algorithm over-relying on the central pixel,it is improved on this basis,and the experimental results obtained are compared with the experimental results of the SGM algorithm.Experiments show that the improved Census transform has a better matching effect than the SGM algorithm,and at the same time the mismatch points are also significantly reduced,and the resulting disparity map is also closer to the standard disparity map.Finally,use the Visual Studio2015 development software combined with the Open CV vision software library to carry out the ranging experiment on the lifting ring.At the same time,in order to avoid the contingency of the experiment,this article observes and analyzes the error of the measured distance by controlling the different distances from the ring to the camera,and analyzes the cause of the error according to the change of the error.The experimental results show that the error of the algorithm within 3 meters is controlled within 1%,which meets the practical application requirements of the text. |
