基于YUV颜色空间码本模型的水下运动目标检测  

Underwater moving targets detection in YUV color space

作　　者：雷飞[1] 黄文路[1] 王雪丽[1] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100124

出　　处：《计算机与应用化学》2014年第4期416-420,共5页Computers and Applied Chemistry

基　　金：2012年科研项目(002000546312507)

摘　　要：水下是1个复杂的动态场景,会受到光照、水波和倒影等影响,同时考虑到水下视频智能监控的实时性的要求,由此该文提出了1种快速有效的运动目标检测算法——基于YUV颜色空间的码本模型。经典的码本模型算法已经能够快速有效的提取出运动目标,但由于该算法在RGB颜色空间下根据颜色扭曲度和亮度范围对每个像素建立码本,计算复杂,运算时间较长,对水波、气泡的抑制较弱,同时无法对阴影进行有效的清除。该文提出的算法根据阴影只改变亮度,对色度影响较小的特点,以及YUV颜色空间的亮度和色度的分离特性,在YUV颜色空间下改进颜色扭曲度和亮度范围的评价方法,然后对视频图像序列建立码本模型。实验表明通过对码本模型算法的改进,对于水下场景,该文算法能够更加完整的提取出运动目标,对阴影、气泡、水波都有较好的抑制,同时算法的实时性也有了很大的提高。Underwater is a complex dynamic scene subjecting to interference such as illumination, water wave, shadow. In order to meet the real-time requirements of underwater video intelligent monitoring system, the codebook model algorithm in YUV color space is presented, which is a fast and efficient moving target detection method. Classic codebook model creates codebook for each pixel according to matching conditions （color distortion and brightness distortion） in RGB color space. It has complicated calculation, longer running time and the weaker the restraint of water wave, bubble and shadow. According to the principle of only change the brightness in the action of shadow as well as luminance and chrominance separation characteristics of YUV color space, algorithm of this paper improves the color distortion and brightness range of evaluation method in the YUV color space and establishes new codebook model for the video image sequence. Experiments show that improvement of codebook model can more complete to extract the moving object, has good inhibition on shadow, bubble, water waves, and has the real-time performance.

关 键 词：码本模型 YUV颜色空间 颜色扭曲度 亮度范围 水下运动目标检测 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]