Research On Wavelet Edge Detection Applied In Visual Image Analysis

Intelligent robot plays a key role in improving production efficiency in industrial production.Introducing machine vision technology into intelligent robot,will improve the operational flexibility and operational complexity.Machine vision images contain a lot of information,and the intelligent robots only need the image edge features,therefore,the core of visual image analysis is image edge extraction,and target pose recognition.Taking the robot as the application background,firstly,the paper compares the commonly used image edge detection algorithms and analyzes their advantages and disadvantages.On the basis,the wavelet edge detection method is proposed and the quadratic B-spline function is chosen as wavelet function to deal with the image taken from the packaging box.Due to the impact of the actual work environment,the machine vision system image noise is bigger,the effect of wavelet transform filtering image noise is the best.Secondly,aiming at partial edge loss and weak edge problem in the wavelet edge detection image,an improved image edge detection method based on wavelet transform is proposed,in this method the image edge detection algorithm wavelet transform and mathematical morphology are combined.On the basis of edge detection,the paper also researches on the image recognition algorithm,through the coordinate transformation,calculates the space position and pose of the box in the machine vision system,accurately extracts the box pose information.Finally,building the machine vision detection system,carries out the analysis and selection of the hardware used in the system.To develop the Motoman robot control platform and image processing unit of robot vision detection system based on VC++.By collecting visual image and image analysis processing,it accurately extracts the pose information of the box,and provides accurate visual image information to control the robot handling box.Edge detection algorithm based on wavelet transform and mathematical morphology and image recognition,can provide accurate box pose information,and it is significant to machine vision system using for intelligent transportation of packing box.