Workshop Workpiece Recognition Based On Binocular Stereo Vision

Today,the manufacturing industry is facing the challenges of increasing labor costs and shortening the product life cycle.Especially in the heavy industry,workers have high labor intensity,high workshop temperatures,and dangerous operations.Machine vision is required to replace manual identification of workshop workpieces and obtain materials.Three-dimensional information to facilitate the machine to grab materials.At present,the cost of machine vision is still relatively high,especially some small and medium-sized enterprises are reluctant to invest a lot of money to introduce vision systems.Moreover,in machine vision,accuracy and cost,and recognition time are often contrary.The higher the accuracy,the recognition The longer the time,the higher the cost.In response to this,this paper has developed a recognition method for quickly acquiring the three-dimensional information of the workpiece according to the actual situation of the manufacturing workshop.This method greatly shortens the recognition time and reduces the cost of machine vision under the requirements of the actual workshop accuracy.Combining reality provides a solution.The research method is as follows:First,the binocular camera is calibrated.The purpose of the calibration is to find a suitable camera model,through which the two-dimensional coordinate system in the image is converted into the three-dimensional coordinate system in space.And through the MATLAB toolbox and NI toolbox to calibrate the binocular camera at the same time,get two camera models,and then through experiments to verify which camera model has good stability and high accuracy,this camera model is selected as the basis for subsequent reconstruction.Then,identify the acquired workpiece image.The purpose of the identification is to find the target object we need in the image.The second is to reduce the reconstruction range and shorten the processing time.Before identification,preprocess the workpiece image.Make the contrast between the workpiece area and the background area stronger and easier to identify,and then use geometric matching to identify the workpiece,and according to the actual workshop situation,it is found that the workpiece can not use the existing geometric matching in the case of reflective and mutual occlusion.Fully identify all the workpieces,analyze the reasons for this situation,and finally propose an improved geometric matching method by averaging multiple templates.Through experiments,this method is even when the workpieces reflect and block each other.Can identify all workpieces.Finally,edge extraction is performed on the identified workpieces.Edge extraction can further reduce the reconstruction range and shorten the processing time,and then analyze the edge contour.It is found that most of the points in the edge contour do not provide effective geometric information,only a few Special points can represent most of the information of the shape of the workpiece,so the extraction of these special points can shorten the processing time and speed up the recognition speed on the basis of not affecting the recognition of the three-dimensional information of the workpiece.After the special points are extracted,the special points in the left and right images are matched one-to-one through the feature-based geometric matching algorithm,and the extra special points are eliminated.Finally,the camera model is used to reconstruct the special point pairs corresponding to the left and right images to obtain The world coordinate values are used to calculate the three-dimensional size information of the workpiece,so as to realize the recognition of the three-dimensional information of the workpiece.From the pre-processing of the workpiece image to the final reconstruction of the special point,the entire process took only 1.63 s.Compared with the general reconstruction,the processing time is shortened by at least 6 times,and the calculated three-dimensional size information of the workpiece also meets The accuracy requirements of the actual workshop.In addition,after the three-dimensional information recognition of the workpiece,the posture of the vehicle when carrying the workpiece in the workshop was simply studied,and a method of vehicle posture correction was extracted through the relative positional relationship between the vehicle and the area to be commanded.