Research On Restoration And Denoising Of Gm-APD Array Laser Radar Range Image

Gm-APD(Geiger-Mode of Avalanche Photodiodes)has become one of the research focuses in the field of laser radar because of its single photon sensitivity,which makes remote target detection possible.The property of Gm-APD array radar range image is very different from the traditional range image,which results in the inability to use conventional range image processing techniques.This dissertation improved the properties of range image by the Gm-APD array radar range image restoration technology and noise suppression technology.Firstly,the working principle of the Gm-APD array laser radar system is introduced in this dissertation.The noise of the system and the measured range image characteristics of the Gm-APD array laser radar are analyzed.The simulation range images is obtained by the simulation of the Gm-APD array laser radar,which is used for the subsequent proposed and verification of image processing algorithm.Finally,the noise characteristics of the simulation range image are analyzed,which shows that the range noise is white noise and also lays the foundation for the subsequent image processing algorithm.Secondly,this dissertation proposed a step-by-step restoration method to restore the range images,which uses median filtering,mean filtering,spatial proximity functions,and correlation adaptive filtering.The image can be restored by only four frames of original range images.The results of image quality evaluation showed that the proposed algorithm has a excellent performance on the recovery of simulated range images.The ISNR is 22.66,the RMSE is 68% lower than the noisy range image,and the SSIM rised 20% comparable to the noisy one.A large amount of lost information is recovered and the properties of the simulation range image is greatly improved.Then,for the remaining missing information and range anomalous noise in the restored range image,this dissertation proposed a concave and convex segmentation noise suppression algorithm,and introduced a combination of open and close operations in grayscale morphology to process the restoration range image.The image quality evaluation results showed that the ISNR is 32.75;the RMSE value is reduced by 40% compared to the recovery range image.The SSIM has risen by 17%compared to the restoration range image.It has successfully removed the missing information and range anomalous noise,while it has successfully protected the target edge information.The properties of range image has been further improved.Finally,a 32×32 Gm-APD laser radar imaging experiment platform was set up to perform range imaging experiments to verify the proposed range imagerestoration algorithm and noise suppression algorithm.Experimental results showed that the retults of measured range image processing are almost the same as the simulation range image processing.The processed range images have a series of advantages such as uniform pixel distribution,more complete target area information,more continuous target edges,and no isolated noise points.The properties of range images have been significantly improved.This article also developed a set of properties of the image processing software based on MATLAB,which verifies the processing effects of range restoration algorithm and noise suppression algorithm,and realized the operation visualization and one-step operation of range image processing.