论文部分内容阅读
相干激光成像雷达距离像处理的一个重要内容就是进行背景抑制。利用原始强度像的均值信息进行距离像的背景抑制因为强度像受到噪声影响而效果不佳,改进算法加入了强度像的噪声滤除,大大提高了背景抑制能力。但是这种利用强度像均值的背景抑制算法要求目标区占有较大的面积,对于小目标图像其抑制效果变差。分析了强度像的直方图特征,提出了一种熵阈值分割抑制距离像背景算法,此算法将强度像的直方图划分为描述目标区像素和背景像素的两个概率分布,而将使这些概率分布熵最大的灰度值作为分割阈值。将此算法应用于实际图像处理,结果表明对于大目标图像和小目标图像都有较好的抑制效果。
Coherent laser imaging Radar image processing is an important part of the background suppression. Since the background image of the distance image using the mean value of the original intensity image is not effective because the image intensity is affected by the noise, the improved algorithm adds the noise image of the intensity image and greatly improves the background suppression ability. However, this kind of background suppression algorithm using intensity as mean requires that the target area occupy a large area, and its suppression effect on the small target image becomes worse. This paper analyzes the histogram feature of intensity image and proposes a method of entropy threshold segmentation to suppress the distance image background algorithm. The algorithm divides the histogram of intensity image into two probability distributions describing pixels of target area and background pixels, The gray value with the largest distribution entropy is used as the segmentation threshold. The algorithm is applied to the actual image processing, the results show that for large target images and small target images have better suppression.