融合区域纹理梯度的船舶阴影视频去除算法

借助于监控视频实时提取内河船舶运行流量等参数的船舶检测系统得以广泛应用,而传统的视频检测方法往往难以有效区分船体边界,造成颜色、纹理相似的前景和阴影区域检测相混淆.为此,解析了HSV颜色特征和LBP纹理不变性检测阴影原理,改进SE-CT阴影去除算法,提出基于颜色和纹理梯度特征GA-HT梯度填充的船舶阴影去除算法.测试表明,GA-HT算法能很好地去除船舶阴影,提高船舶的目标识别的精度和实时性,阴影检测综合指标F达到92％.

Ship Shadow Elimination Algorithm Based on HSV and Texture of Gradient Amendment

At present,the monitoring system by means of real-time video has been widely used,which can extract inland parameters such as ship traffic,etc.while the main methods of shadow detection are often difficult to effectively segment the foreground and shadow because of similar color and texture information in the border region.Through analyzing the LBP texture invariance and HSV color feature extraction principle of shadow,a novel ship shadow elimination is proposed.Chromacity information is first used to create a mask of candiate shadow pixels,follwed by employing improved Local Binary Pattern(LBP) in extracting texture information to remove foregroud pixels that were incorrectly included in the mask,and then applying gradient amendent to obtain a ultimate shadow.Experiments indicate that the proposed algorithm can effectively remove the shadow of the ship,improve the ship's Matching Pursuit accuracy and real-time capability,and final shadow detection comprehensive index F is up to 92％.