| 基于小波域特征和贝叶斯估计的目标检测算法 |
| |
| 引用本文： | 刘英霞,王希常,唐晓丽,常发亮. 基于小波域特征和贝叶斯估计的目标检测算法[J]. 山东大学学报(工学版), 2017, 47(2): 63-70. DOI: 10.6040/j.issn.1672-3961.0.2016.174 |
| |
| 作者姓名： | 刘英霞 王希常 唐晓丽 常发亮 |
| |
| 作者单位： | 1. 山东传媒职业学院, 山东 济南 250200;2. 山东省教育招生考试院, 山东 济南 250011;3. 纪念斯隆-凯特琳癌症中心医学物理学部, 纽约 西哈里森10606 美国;4. 山东大学控制科学与工程学院, 山东 济南 250011 |
| |
| 基金项目： | 山东省教育科学“十二五”规划资助项目(ZK1437B015);国家自然科学基金资助项目(60975025,61273277) |
| |
| 摘 要： | 为了改进目标检测算法,在小波域建立基于贝叶斯概率估计的模型,得到一个自适应最佳阈值,并利用该阈值得到待检测的目标。对待检测的图像序列进行基于滑动窗口的双Haar小波变换,对小波变换后的低频分量建立基于核密度函数的贝叶斯概率估计模型,通过训练和学习,得到自适应的最佳阈值,利用该阈值对低频分量进行判别,得到只含有目标的二值化图像。选取室内室外一个和多个运动目标的6个视频序列对该算法的有效性进行检验,并同其他算法相比,可以给出更好的检测结果。
|
| 关 键 词： | 小波域 贝叶斯概率估计 目标检测 动态背景 |
| 收稿时间： | 2016-05-23 |
Object detection algorithm based on Bayesian probability estimation in wavelet domain |
| |
| LIU Yingxia,WANG Xichang,TANG Xiaoli,CHANG Faliang. Object detection algorithm based on Bayesian probability estimation in wavelet domain[J]. Journal of Shandong University of Technology, 2017, 47(2): 63-70. DOI: 10.6040/j.issn.1672-3961.0.2016.174 |
| |
| Authors: | LIU Yingxia WANG Xichang TANG Xiaoli CHANG Faliang |
| |
| Affiliation: | 1. Shandong Communication and Media College, Jinan 250200, Shandong, China;2. Shandong Province Academy of Education Recruitment and Examination, Jinan 250011, Shandong, China;3. Department of Medical Physics, Memorial Sloan Kettering Cancer Center, West Harrison 10606, New York, America;4. School of Control Science and Engineering, Shandong University, Jinan 250011, Shandong, China |
| |
| Abstract: | In order to improve the detection algorithm, Bayesian probability estimation model in wavelet domain was built to get a robust threshold, and the detected object could be obtained with the adaptive threshold. Moving Window-Based Double Haar Wavelet Transform for detected image sequence was finished. Bayesian probobility estimation model based on kernel density function was built for low frequency part, and adaptive threshold could be obtained after training and studying. With the threshold to judge the low frequnency part, the binary image could be got. Six video sequences with one targe and multiple targets outdoor and indoor were employed to evaluate the effectiveness of presented algorithm. Experimental results showed that it could give a better detecting results. |
| |
| Keywords: | Bayesian probability estimation object detection wavelet domain dynamic background |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《山东大学学报(工学版)》浏览原始摘要信息 |
|
点击此处可从《山东大学学报(工学版)》下载全文 |
|
