融合区域纹理梯度的船舶阴影视频去除算法

发布时间：2018-04-28 02:09

本文选题：视频监控 + 图像处理　；参考：《北京邮电大学学报》2017年S1期

【摘要】：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%.
[Abstract]:The ship detection system based on real-time extraction of parameters such as the running flow of inland river ships with the help of surveillance video has been widely used, but the traditional video detection methods are often difficult to distinguish the hull boundaries effectively, resulting in color. Texture similar foreground is confused with shadow area detection. In this paper, the principle of HSV color feature and LBP texture invariance detection is analyzed, the SE-CT shadow removal algorithm is improved, and the ship shadow removal algorithm based on color and texture gradient GA-HT gradient filling is proposed. The test results show that the GA-HT algorithm can remove ship shadow and improve the accuracy and real-time performance of ship target recognition. The comprehensive index F of shadow detection is 92g.
【作者单位】：南京大学电子科学与工程学院;
【基金】：国家自然科学青年基金项目(61502226) 国家船联网重大专项项目(2012-364-641-209)
【分类号】：TP391.41

【相似文献】