激光雷達(dá)距離像背景抑制算法研究

相干激光成像雷達(dá)距離像處理的一個(gè)重要內(nèi)容就是進(jìn)行背景抑制。利用原始強(qiáng)度像的均值信息進(jìn)行距離像的背景抑制因?yàn)閺?qiáng)度像受到噪聲影響而效果不佳,改進(jìn)算法加入了強(qiáng)度像的噪聲濾除,大大提高了背景抑制能力。但是這種利用強(qiáng)度像均值的背景抑制算法要求目標(biāo)區(qū)占有較大的面積,對(duì)于小目標(biāo)圖像其抑制效果變差。分析了強(qiáng)度像的直方圖特征,提出了一種熵閾值分割抑制距離像背景算法,此算法將強(qiáng)度像的直方圖劃分為描述目標(biāo)區(qū)像素和背景像素的兩個(gè)概率分布,而將使這些概率分布熵最大的灰度值作為分割閾值。將此算法應(yīng)用于實(shí)際圖像處理,結(jié)果表明對(duì)于大目標(biāo)圖像和小目標(biāo)圖像都有較好的抑制效果。
One of the most important content s about coherent imaging laser radar range image processing is suppressing the background. If using the primitive meaning information of the intensity image to suppress the background of range image , the effect would not be good due to the influence of noise. The improved algorithm adds noise filter about the intensity image , which greatly improved the background suppression capacity. Also the previous algorithm claims the object to have relatively larger area , which means a worse effect to small object
images. The article analyzes the histogram characteristics of the intensity image , brings forward an algorithm of thresholding segmentation suppression range image background using the ent ropy of the histogram , which divides the intensity′s histogram into two probability dist ributions : object region pixels and background pixels and regards the maximum value of probabilistic dist ribution ent ropy as the threshold of the segmentation. Using this method in the actual image processing , the result reveals that it has good suppressing effect on both small objects and large ones.