Research On The Around View Image System Of Engineering Vehicle

With the rapid development of China's economy,real estate market booms sustainedly,investment of fundamental construction continues,so the demand for engineering vehicles is also increasing.Engineering vehicles play an important role,but also accompanied by a lot of security risks.Since engineering vehicles' operating environment is bad and the bodies are huge,only relying on the rearview mirrors is difficult to ensure the safety.And nowadays the science and technology updates so quickly that more engineering vehicle electronic products emerged,therefore advanced technology applied to traditional engineering vehicle industry is an inevitable trend.Around View Image System can provide the bird's-eye view image around the vehicle real-timely,and effectively eliminate the blind area,which provides a very effective auxiliary effect for the drivers.It is of great practical value to study the system and apply it to the engineering vehicles.This paper studied the background and research status of vehicle around view system,and deeply analyzed the key technology of this system,as a result,a practical solution was put forward.The system uses six fish-eye cameras to acquire images around the vehicle and the images are distorted,so the first step is to correct the distortion,then transform the corrected images into bird's-eye view images using perspective transformation,and finally the six images are stitched and fused.In the algorithm of fisheye lens distortion correction,this paper studied several commonly used correction algorithm,then compared the effects through the simulation experiment,and finally selected calibration method as the correction algorithm.The calibration principle and method were introduced in detail,and the calibration of the camera was realized step by step.The experimental result shows that calibration method meets the requirements of the system.In the section of bird's-eye view transformation,the transformation model was analyzed,and the scheme that calculating the transformation matrix by finding corner was confirmed.Besides,the method of selecting the reference points was given.In the image matching algorithm,the principle and realization method of SIFT algorithm was introduced.Through researching and improving the traditional SIFT algorithm,an improved SIFT algorithm based on part feature points was put forward.The improved algorithm extracts featured points in overlap regions only,which can greatly reduce the number of feature points and operation time,and successfully improve the rate and efficiency in the same time.The method of identifying the overlap region was given in this paper and the effectiveness of the improved algorithm was verified by comparison of experimental results.In image fusion algorithm,this paper studied several common pixel level fusion algorithms and selected the Fade In-out fusion algorithm based on the simulation comparison.Fade In-out fusion algorithm can't completely eliminate the seam in the condition that color value of the two images is widely different and the transition is not natural.Based on the shortcomings,this paper improved the Fade In-out fusion algorithm.The improved algorithm calculates and compares the average gray value of overlap region and fuse the images after adjusting gray value to similar.The experiment results show that the improved algorithm can effectively eliminate the mosaic traces and make fused image more natural.This paper studied the algorithms of vehicle around view system,and made some improvement of fisheye lens distortion correction algorithm,image matching algorithm and image fusion algorithm,which improves the visual effect of around view image and effectively reduces the complexity of the algorithms.All of this has good application prospect and value.