Research On High-speed Process Imaging Acquisition Based On Compressed Sensing In Vehicle Engineering

As a new direction in the field of image processing,compressed sensing theory has been widely concerned by many scholars since its appearance.In the field of high-speed imaging in vehicle engineering,the use of compressed sensing theory can break through the traditional imaging methods,the measurement matrix was difficult to implement by hardware and the imaging quality was not high.This paper focuses on these two issues,and makes an in-depth analysis and research on the measurement matrix and the optimization of the measurement matrix used in the high-speed process industrial imaging system in vehicle engineering.Specifically,the main work of this article is as follows:(1)The compressed sensing theory is explained,the calculation formula of the industrial imaging system based on the compressed sensing theory is deduced,the measurement matrix and the matrix optimization method used are simulated and analyzed,and the differences in the industrial imaging system are compared.Measure the matrix and the signal reconstruction effect before and after matrix optimization.(2)Using the 0-1 matrix as the measurement matrix can solve the problem that the measurement matrix was difficult to implement in hardware,the 0-1 circulant matrix and the 0-1 random matrix are respectively used as the measurement matrix in the industrial imaging system,and the sparse signal and the compressible signal are collected and reconstructed respectively.The effects of different reconstruction matrices,optimized matrices and approximate matrices on sparse signals and compressible signals are compared and analyzed.The simulation results show that: 0-1 cyclic matrix and 0-1random matrix can be used to collect sparse signals;when collecting compressible signals and using discrete cosine transform matrix as sparse base,0-1 random matrix can be used instead of 0-1 For the circulant matrix,the 0-1 circulant matrix is used as the measurement matrix and the reconstruction effect is not good.(3)There are few studies on the influence of the change of the internal elements of the 0-1 matrix on the performance of the reconstruction matrix.For this problem,the shift step length of element 1 in the 0-1 circulant matrix is changed.The simulation results show that the reconstruction matrix,optimized matrix and approximate matrix based on the changed 0-1 circulant matrix can reconstruct the signal.There is a significant improvement.The number of rows and columns in the 0-1 random matrix is changed,and the simulation results show that the reconstruction matrix,optimized matrix and approximate matrix obtained based on the changed 0-1 random matrix can significantly improve the signal reconstruction performance.(4)Aiming at the blindness of measurement matrix selection in industrial imaging systems,a method of screening measurement matrices through contour maps is proposed.Statistics and draw the 0-1 circulant matrix with the maximum number of columns of 256 and 4096,the contour map and the difference map of the 0-1 random matrix and the Gaussian matrix before and after optimization,and the 0-1 circulant matrix with the maximum number of columns of 4096,and the approximate matrix obtained based on the matrix is brought into the industrial imaging system for experiments.The simulation results show that the large-dimensional 0-1 cyclic matrix used as the measurement matrix in the industrial imaging system can collect more effective information of the observed object,and can image and measure the high-speed process in vehicle engineering in a short time.After the matrix is optimized,the reconstruction effect is improved.Using the optimized matrix to collect and reconstruct the signal can solve the problem of low imaging quality in industrial imaging systems.Through the research of this article,it provides theoretical and technical support for the measurement matrix and the optimization of the measurement matrix used in the high-speed process industrial imaging system in vehicle engineering.