Research On Image Measurement Technique Based On Binocular Stereo Vision

With the rapid development of computer vision technology,image processing technology based on binocular stereo vision has been paid more and more attention in the field of object size measurement in recent years.At present,binocular vision measurement technology is mostly limited to the application of close-range and small-size object,while the measurement of large-size object is rarely studied.In the actual measurement environment,the traditional manual contact measurement method has the disadvantages of low efficiency,poor operability,and low measurement accuracy for object with large size and high position.Based on this,this paper focuses on the actual demand and mainly studies the application of binocular stereo vision technology in large building size measurement,including camera calibration,epipolar rectification,image preprocessing,stereo matching and 3-D information recovery in binocular image measurement,and finally realizes the size measurement of large building.The main work of this paper is as follows:1.Camera calibration is the primary task of image measurement.In this paper,a camera calibration method based on plane template with strong operability and high calibration accuracy is adopted to complete the calibration of binocular camera.On the basis of obtaining calibration parameters,aiming at the problem of the nonparallel epipolar in the binocular image,a Bouguet epipolar rectification algorithm is proposed to realize stereo rectification of binocular image.2.Aiming at the characteristics of the building image captured in the outdoor environment,a set of image preprocessing methods suitable for the experimental image is designed.To solve the problem of image noise pollution,a 2-D wavelet threshold denoising algorithm is proposed.Aiming at the problem of uneven distribution of image brightness,the improved histogram equalization algorithm is used to improve the image quality by effectively adjusting the overall contrast of the image,which provides a guarantee for the subsequent feature point matching.3.The quality of stereo matching algorithm directly affects the accuracy of the final binocular image measurement results.In this paper,the principle of SIFT feature matching algorithm is deeply studied.To solve the problems of repeated matching and mismatching of feature points in the matching process,the improved SIFT feature matching algorithm based on feature vector bidirectional matching and RACSAC is applied to realize the stereo matching of building image.4.The realization scheme of object size measurement based on binocular stereo vision technology is proposed.On the basis of analyzing the feasibility of the existing binocular vision measurement algorithm,the baseline distance parameter can be adjusted to make it applicable to the measurement of large object.Besides,buildings in different scenarios are tested.Finally,we using Auto CAD to complete the building plan drawing and dimensioning according to the experimental measurement data.In this paper,two identical ordinary USB cameras are used to build a binocular image acquisition device,and the experiment is completed under the MATLAB and VS2012 compiler environment.The experimental results show that the image measurement algorithm based on binocular stereo vision can effectively be used to measure the size of large buildings,which has certain practical value.