摘要
为解决运动目标检测算法鲁棒性和实时性差的问题,在原始codebook算法的基础上提出一种改进的codebook算法。在匹配码字时将最近更新的码字调整至码本列表的最前端,加快码字匹配的速度;以适应光照变化且运算简单的局部二值模式(local binary patterns,LBP)直方图向量代替原有的RGB向量,采用码本记录局部区域的纹理特性,并通过实验比较原始的codebook、混合高斯算法及改进后的codebook。结果表明:改进后的codebook算法较其他2种算法具有更快的处理速度和更好的检测效果,且增强对场景中光照变化的适应力。
In order to solve the problem o f robustness and poor real-time performance o f general moving targetdetection algorithm, an improved codebook algorithm is designed based on the original codebook. Adjust the recentlyupdated code word to the most front end o f the codebook list when matching code word in order to improve the speed ofcode marching. Use LBP histogram vector which is adapted to the change o f illumination and simply operated instead o f theoriginal RGB vector. Record the local texture information through codebook, and compare the original codebook withmixed Gaussian algorithm and improve codebook by test. The experiment show that improved codebook has faster speedand better detection results, and stronger adaptability to illumination change in the scene.
作者
张小正
周鑫
袁锁中
王从庆
Zhang Xiaozheng, Zhou X in , Yuan Suozhong, Wang Congqing(College of Automation Engineering, Nanjing University of Aeronautics S: Astronautics, Nanjing 210016, Chin
出处
《兵工自动化》
2018年第2期48-53,共6页
Ordnance Industry Automation
基金
国家自然科学基金(61273050
61573185)