成像激光雷达距离像乘性噪声的分析和处理

发布时间：2018-01-10 11:24

本文关键词：成像激光雷达距离像乘性噪声的分析和处理　出处：《哈尔滨工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：激光雷达探测技术是现阶段国内外研究的热门领域，，在航天、气象、医疗、工业、预警、导弹制导和民用探测等诸多方面都有广泛的应用。成像激光雷达不仅可以获得探测目标的强度像，还可以获得探测目标的距离(3D)像。但是图像往往伴随着各种噪声，尤其是乘性噪声，由于和原始信号相关，很难去除，影响测距精度和成像质量。国内的激光雷达研究起步较晚，目前针对成像激光雷达乘性噪声的分析和处理方法并不系统。本文对基于ICCD增益调制成像激光雷达中乘性噪声的产生原因进行分析，分别从空域和频域角度系统的给出噪声处理方法，并对每种方法的雷达图像处理结果进行对比评价，分析优劣。 ICCD增益调制成像激光雷达的乘性噪声来自成像器件的噪声，经过微通道板的放大作用，以额外增益的形式和图像信号关联在一起。在空域角度利用均值滤波法、中值滤波法和同态滤波法处理雷达图像，同态滤波法对乘性噪声的处理最好，但是滤波时容易受到加性噪声的干扰，所以根据每种滤波法的特点，结合多种滤波法对图像处理可以得到更佳的除噪效果。在频域角度利用傅里叶变换将图像变到频谱，采用低通、带通和多尺度小波变换滤波法处理雷达图像，由于多尺度小波变换对图像细节的辨析能力很强，小波变换滤波在保留边缘的同时可以有效滤除乘性噪声。当雷达图像同时含有乘性和加性噪声时，需要对两种噪声进行并行处理。利用图像分割将探测目标分离出来，对于目标采用同态滤波法去除乘性噪声，对于背景采用中值滤波法滤波去除加性噪声，防止两种噪声互相干扰。利用邻域像素灰度值的涨落来判断乘性噪声，根据中心像素和周围四领域像素的梯度变化可以对加性和乘性噪声加以区分，不同情况使用不同方法进行滤波。在此基础上提出一种超前的六邻域图像的构造和处理方法，可以更充分的利用邻域信息对噪声类型进行判别，得到更好的滤波效果。
[Abstract]:The laser radar detection technology is a hot field of current domestic and international research in aerospace, meteorology, medical, industrial, early warning, missile guidance and civil detection etc. are widely used. The imaging laser radar can not only obtain the target intensity image, also can get the target distance (3D) but like. The image is often accompanied by various kinds of noises, especially due to the multiplicative noise, and the original signal, it is difficult to remove, influence range precision and imaging quality. The domestic research started later analysis and processing of laser radar, laser radar imaging method for the multiplicative noise is not systematic. This paper analyzes the causes of multiplicative noise ICCD gain modulation imaging laser radar based on the noise processing method are given respectively from the perspective of spatial and frequency domain of the system, and for each method of radar image processing results are comparative evaluation The analysis is good and bad.
The noise multiplicative noise from the imaging device ICCD gain modulation imaging laser radar, after amplification of micro channel plate, in the form of additional gain and image signal correlation together. Using the mean filter method in the spatial perspective, median filtering method and homomorphic filtering method of radar image processing, processing of multiplicative noise homomorphic filtering the best, but when the filter is vulnerable to additive noise, so according to the characteristics of each kind of filtering method, the effect of noise removal combined with a variety of filtering method for image processing can get better. In the frequency domain using Fourier transform to the image spectrum, using low-pass, band-pass and multi-scale wavelet transform filtering method of radar the image, due to the multiscale wavelet transform of the image details discrimination ability is very strong, the wavelet transform filter in preserving the edge can effectively filter out the multiplicative noise at the same time. When at the same time with the radar image A multiplicative and additive noise, the need for two kinds of noise for parallel processing. The target is separated by using image segmentation, the target by homomorphic filtering to remove multiplicative noise, the median filtering method for background filter for removing additive noise, prevent mutual interference. Two kinds of noise fluctuation using pixel neighborhood value to determine the multiplicative noise, according to the change of gradient pixels and the surrounding four fields can distinguish between additive and multiplicative noise, different conditions using different filter method. Based on the proposed method and ahead of the six neighborhood image, using neighborhood information to determine the type of noise well, better filtering effect.

【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN958.98

【参考文献】