摘要
针对视频监测区的特定视觉特征进行快速定位,在提高视频监测区的视觉目标定位方面具有重要意义。由于视频监测区一般设置在人流量大且多,使得监测区的目标特征不容易固定。传统的定位方法主要根据目标特征进行定位,当监测区内出现的大量类似的目标特征时不能及时进行检测筛选,导致定位准确度低的问题。提出视频监测区的特定视觉特征快速定位方法研究。通过混合高斯背景模型构建视频监测区多态散布模型,对监测区域的视频图像进行去噪处理,并提取视频图像颜色特征,利用颜色直方图Euler距离实现特定图像特征的跟踪匹配,实现对视觉检测区域的特定视觉特征快速识别定位。实验结果表明,改进的定位方法相比传统算法,对相同视频片段的1000幅图像进行特定视觉特征快速定位时,定位准确度高,时间短。
The fast locating of the specific visual features in video monitoring area is of great significance to im- prove the visual target location of the video monitoring area. A fast locating method for the specific visual features in video monitoring area is proposed. Based on the mixed Gauss background model, a polymorphism distribution model in video monitoring area is built, and the video image in the monitoring area is processed, and the color feature of the video image is extracted. The Euler distance of color histogram is used to realize the tracking and matching of specific image features, and the fast recognition and locating of the specific visual features in the visual monitoring area are realized. The experimental results show that, compared with the traditional algorithm, the improved locating method is of high locating accuracy and short time for the fast locating of the specific visual features of 1000 images in the same video clips.
出处
《计算机仿真》
CSCD
北大核心
2016年第3期355-358,共4页
Computer Simulation
关键词
视频监测区
特定视觉特征
快速定位
Video monitoring area
Specific visual features
fast locating
