视频监控图像去抖动视觉监测算法优化仿真

发布时间：2018-05-03 05:33

本文选题：视频监控 + 图像去抖　；参考：《计算机仿真》2017年09期

【摘要】：对视频监控图像的去抖动视觉监测算法的研究,可更好的对存在尺度、旋转变换的监控抖动视频图像进行去抖处理。对视频图像进行去抖动处理,需要先计算出图像特征点之间的校正量,对视频进行补偿,完成对图像的去抖动视觉监测。传统方法依据像素灰度差值搜索最优匹配项,求解当前标准帧的偏移量,但忽略了对图像特征点的校正,导致视频图像去抖动效果不理想。提出基于卡尔曼滤波的视频监控图像去抖动视觉监测算法。提取视频监控图像的特征点,结合快速近似最邻近库理论匹配视频邻帧间的图像特征点,剔除错误匹配的图像特征点,结合卡尔曼滤波理论对视频监控图像特征点轨迹进行滤波处理,计算出滤波前后图像特征点之间的校正量,直接对监控视频进行补偿,消除监控视频抖动。实验结果表明,所提算法能够有效减轻视频监控图像的抖动现象,具有更强的实时性及健壮性。
[Abstract]:The research on the de-jitter visual monitoring algorithm of the video surveillance image can better deal with the jitter video image with the existing scale and rotation transformation. In order to dejitter the video image, it is necessary to calculate the correction amount between the feature points of the image, compensate the video, and complete the visual monitoring of the image de-jitter. The traditional method is based on the pixel grayscale difference to search the optimal matching term to solve the offset of the current standard frame, but the correction of the image feature points is neglected, resulting in the video image de-jitter effect is not ideal. A visual monitoring algorithm for video surveillance image dejitter based on Kalman filter is proposed. The feature points of video surveillance images are extracted, and the image feature points between adjacent frames are matched with the theory of fast approximation nearest nearest library, and the image feature points that are mismatched are eliminated. Based on Kalman filter theory, the trace of feature points of video surveillance image is filtered, and the correction between the feature points of the image before and after filtering is calculated, which can compensate the video directly and eliminate the jitter of the video. The experimental results show that the proposed algorithm can effectively reduce the jitter of video surveillance images and has better real-time and robustness.
【作者单位】：天津商业大学宝德学院;
【分类号】：TN948.6;TP391.41

【相似文献】