| Compared with microwave radar, laser radar has better concealment, higher signal to noise ratio and resolution ratio,and it has developed quickly in the last few years. But there is big difference in signal recording technology and way. And now there are two mainstream ways. One is simulated count technology, and the other is single photon technique. Single photon technology has attracted much attention due to its higher sensitivity and better resolution. Time-correlated single photon counting(TCSPC) technique was applied in fluorescence lifetime testing earliest, which has a better time resolution and more higher time sensitivity. Therefore, if this technique is applied to single photon laser radar, the whole performance of the radar system will reach a new level.According to the work principle of laser radar, this paper selects appropriate detectors, building a single photon laser radar system based on time-correlated single photon counting technique, and obtains 3D point cloud data by scanning the target.But due to the system hardware and environmental conditions, the point cloud data which single laser radar system obtains must contain noise, jitter and other factors that affecting image quality. Through studying, this paper achieved denoising processing using the mobile surface fitting method and mathematical morphology filtering algorithm, and quantitatively analyzed the effect of the two methods. The target is assumed as the plane target, the curved surface target and the step type target. Considering the advantages and disadvantares of the two methods and statistic characteristics of point cloud data, this paper has proposed improved moving surface fitting denoising algorithm based on mathematical morphology, establishing a new denoising model of "first to denoise, then hole-filling, next iteration". Using the improved algorithm, the paper processed the curved surface target, comparing and analying the iterative times and the experimental error. |
PDF Full Text Request