Design And Implementation Of Panorama Image Splicing Algorithm In Parking System

With the rapid development of economy in domestic,automobile has been used as transportation tools in thousands of households.However,with the increasing of automobile,parking problems begin to appear in many cities.Furthermore,the horizon is limited with the traditional rearview mirror.Thus,scratch incidents or even serious traffic accidents increased.In order to solve auxiliary parking,the panoramic parking system appeared,which inputted the pictorial information surrounding the car by cameras.With the appropriate image processing method,a top view of the environment surrounding the car is obtained,which provides effective information around the car for automatic parking of the car.Based on studying the panoramic image Mosaic algorithm of parking system and analyzing algorithm optimization strategy,we proposed a panoramic parking system implementation scheme.Fisheye cameras were installed in four sides of the car,and then verified imagines from the four directions of fisheye cameras and converted into aerial view,which would be carried out by an adaptive panoramic splicing method,and eliminate stitching line by image fusion algorithm.Based on MATLAB 8.3,the algorithm is practiced and verified.By comparing the display effect of the panoramic top view obtained by the software model with the actual panoramic top view,the feasibility of the panoramic image stitching algorithm of the panoramic parking system is proved.Based on study of panoramic parking system,the hardware acceleration of panoramic parking system is designed and practiced.According to the algorithm process module partition,various of modules of the panoramic parking system algorithm were implemented as hardware IP by Verilog.Finally,module integration was carried out,and conducted joint simulation by ISE14.7 and Modelsim SE-64-10.2 – c.Compared with the pixel coordinate data obtained by the software algorithm model,the result of final pixel coordinates data confirmed the accuracy of the hardware implementation in which Euclidean distance between the coordinates was taken as the error value,and the error was controlled within 0.274 cm.Finally,the hardware algorithm is transplanted into the Xilinx spant6-xc6slx45.Parameter calibration is carried out in the real vehicle scene,and the final stitchingimages and the original image in all directions were verified through the LCD screen with a resolution of 1920*720,which assessed the effect of actual display based on image quality.Thus,the feasibility and value of the hardware system were confirmed in this study.