Research On Image Enhancement Method In Panorama Parking Assist System

With the growing popularity of the vehicle, the traffic congestion have become increasingly serious problem, which is a huge threat to people and driving. Lately, for the sake of reduce the road accidents, there are many car types installed with panorama parking assist system, and it is very convenient to reversing and parking. Panorama parking assist system is consisted of four cameras that can collect video images around the vehicle, and composite image from cameras all showed in LED screen. But for uneven illumination and variable driving environment, resulting in different brightness and contrast, so it is a trouble for display device and viewing. So as to retain consistent illumination image, it is necessary to its influence. What’s more, In order to eliminate noise from image acquisition, transmission and enhancement process, this paper put forward multi-innovation Kalman filtering based on standard Kalman filtering. For image enhancement algorithms, the followed researches in this paper are referred.First, this section introduce the basic theory of image enhancement and analyze the their advantages and disadvantages of the traditional image enhancement methods. Especially for the principle and process of the histogram equalization algorithm.Second, the basic theory is analyzed for Retinex algorithm, and carefully studied the different methods based on improved Retinex algorithm, and mainly researched center / surround Retinex algorithm(including SSR、MSR、MSRCR).Next, the feature of image in the panoramic parking assist system is analyzed, the paper combine HE with MSR method to deal input image. Through simulation experiments, the result of this paper is more better than HE, SSR and MSR enhancement algorithm. By subjective evaluation and objective evaluation method, results of this combined algorithm has a lower time complexity, while the mean, variance, entropy, MSE, peak signal to noise ratio has a better improvement.At last, the image model based on MFR is build and standard Kalman filtering is introduced. At the same time, this paper proposes multi-innovation Kalman filter and verifies the advantage of this method by simulation. The simulation results show that enhanced image retains less noise and more details.