一种基于SSIM的小波域离散余弦变换多车目标识别算法  

Algorithm for multi-vehicle target recognition based on discrete cosine transform in wavelet domain and SSIM

作　　者：蔡黎[1] 代妮娜[1] 刘龙成 戴闽鲁[1,3] 王欣煜 CAI Li;DAI Nina;LIU Longcheng;DAI Minlu;WANG Xinyu(Signal and information processing Key Lab, Chongqing Three Gorges University, Chongqing 404000, China;Sweden Royal Institute of Technology, Stockholm 10512, Sweden;Beijing Digital Television National Engineering Laboratory, Beijing 101000, China)

机构地区：[1]重庆三峡学院信息与信号处理重点实验室,重庆404000 [2]瑞典皇家理工学院,瑞典10512 [3]数字电视国家工程实验室(北京),北京101000

出　　处：《电视技术》2017年第11期184-186,202,共4页Video Engineering

基　　金：教育部"春晖计划"资助项目(Z2015138);重庆市教委科学技术研究项目(KJ1501027);重庆三峡学院青年项目(16QN06)

摘　　要：针对车辆目标识别算法在多车目标情况下识别效果较差,识别时间长的问题,提出一种基于SSIM的小波域离散余弦变换多车目标识别算法。考虑现有文献算法相关环节存在的不足,对其进行了优化和改进。新算法在对运动车辆进行目标识别时,增加结构相似度判别环节,进行离散余弦变换,并引入小波域,通过处理、计算图像的形态特征值,最终以矩阵的形式得到识别结果。实验表明,该算法在多车目标情况下,能够大幅度提高车辆目标识别率,并在一定程度上缩短识别时间,性能优于现有算法。Aiming at the problem that the target recognition effect is poor and the recognition time is long when the vehicle target recognition algorithm is used in the presence of multiple vehicles,Algorithm for Multi-Vehicle Target Recognition Based on Discrete Cosine Transform in Wavelet Domain and SSIM is presented.Considering the deficiency of the traditional algorithm,the optimization and improvement are carried out.When the new algorithm is used to identify the moving vehicle,the structure similarity is added,the discrete cosine transform is used for wavelet domain is introduced in the algorithm.By processing and calculating the morphological features of the image,the final results are obtained in the form of matrix.The experimental results show that in the case of multiple vehicles,the proposed algorithm can greatly improve the vehicle target recognition rate by a large margin and reduce the recognition time to a certain extent,and the performance is better than the existing algorithms.

关 键 词：目标识别 结构相似度方法 离散余弦变换 小波域 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]