Size Detection Of Automobile Front Frame Based On Machine Vision

As a part of the automobile chassis,the front subframe needs to support other suspension brackets in the vehicle,which plays an important role in the maneuverability and comfort for the car.Therefore,the size detection was important to ensure the good performance for the car.Visual image inspection technology,which has the advantages of high precision,high speed and non-contact,has been widely used in size detection and surface quality inspection of parts.The front subframe was large and there were some geometric features unevenly distributed in the surface.Due to the presence of low accuracy feature matching,the dimension measurement accuracy was not high for the size inspection of large and complex castings.Therefore,take the front subframe as the testing object,the size measurement was researched for large and complex castings.The main research contents are as follows:(1)According to the characteristics of the front subframe,the flip frame and three-axis control system were designed,and the image acquisition unit was selected,the construction of the image acquisition test-bed was completed.At the same time,in order to eliminate image distortion,Zhang Zhengyou checkerboard calibration method was used to calibrate the acquisition system for improve the accuracy of subsequent image detection.(2)A Hu moment invariant template matching algorithm combined with fuzzy comprehensive evaluation method was proposed.Firstly,the image was converted into a binary image by image preprocessing.Then,the fuzzy membership function was used to process the geometric feature images and established standard template library for front subframe.Then the eigenvalues were extracted by Hu invariant moment for the preprocessed binary image and template library image,and the search strategy and similarity measurement function were defined to locate and recognize the geometric features in the front subframe surface.(3)The geometric features were extracted separately by defining the region of interest for the front subframe surface.The Harris algorithm was used to detect the corners to obtain the candidate corners for geometric feature.In order to obtain the key corner points of the image,the Susan algorithm was used for secondary detection to obtain the key corners,which was convenient for the subsequent size measurement.(4)Through the design of GUI interface program,the front subframe template matching algorithm and corner detection method were integrated to complete the image processing.In addition,the geometric feature size was calculated,and compared with the actual size for the front subframe.Furthermore,the error analysis shows that the above two algorithms can meet the requirements of dimension measurement error.