Irregular Materials Volume Measurement Base On Binocular Vision

With the advances of industrialization in recent years,irregular materials’ volume measurement has become a popular research direction,which is mainly realized by binocular stereo vision technology.However,the current volume measurement requires professional software hardware and complex calculation,therefore they cannot meet production throughput requirements.In view of above problems,this research presents a method of irregular volume measurement based on binocular stereo vision.The research approaches are:(1)Introducing a secondary calibration method that is based on Zhang’s camera calibration theory using MATLAB toolbox.The original calibration requires high quality images and human intervention to select specific points(e.g.corners)in order to obtain more accurate calibration results.The repeated calibration,manpower consumption,and easy to accumulate error are not preferred.For improvements,the initial calibration data set with the minimum re-projection error is selected to conduct the secondary calibration in order to simplify the calibration process and optimize the calibration results.(2)Image denoising.To minimize effects of noises in the image collected by USB camera module,median filtering and histogram equalizing methods are used before the stereo matching algorithm based on Semi-Global Block Matching applied to obtain the parallax information of left and right images.This approach can effectively suppress the image noise and enhance the details of the image features and reduce the false matching rate of subsequent stereo matching.(3)Investigating a suitable stereo matching algorithm.According to the results of secondary calibration,the pole-line correction of image pair is carried out.Stereo matching algorithm based on Semi-Global Block Matching is introduced to obtain parallax information of left and right image pairs.(4)Applying a volume measurement method to the target area.The Cloud Compare software is used to segment the scene information,extract the target area,and follow by 3D reconstruction and volume calculation of the target area to reduce the amount of calculation.This method eliminate the step of taking the pre-loaded scene as a reference,and reduce complex and error-sensitive calculations.(5)Designing and building an experimental platform for the irregular volume measurement.The experimental results show that the methodology used in this study meets the measurement accuracy of 10% or less,and provides a strong theoretical basis and reference for the subsequent practical application development.