基于代数多重网格的图像拼接中动态目标的检测

Dynamic Object Detection in Image Mosaic Based on Algebraic Multigrid

摘要: 图像拼接中出现的运动目标可能使拼接出现不能正常拼接或者拼接出多重影像的现象。本文提出一种图像拼接的运动目标检测方法，去除运动目标对图像拼接的影响。首先将采集的多场景图像进行拼接，如果不能正常拼接，则利用代数多重网格(AMG)方法对原始图像进行处理，得到重构的多层网格图像，在此基础上利用背景差分法对运动目标进行检测消除干扰。如拼接中出现多重影像，使用上述过程提取运动目标，分析运动目标的轨迹。实验结果表明，本文算法具有较强的鲁棒性，能有效的去除噪声的干扰和准确地检测出运动目标。

Abstract: The moving targets appearing in image mosaic may cause results of the abnormal splicing or mul-tiple images of the same object appearing in the result image. This paper presents a method of de-tection of moving target in image mosaic, so as to remove the effect of moving targets in the image mosaic. Multi-scene images collected should be spliced firstly. If the splice procedure cannot be processed normally, algebraic multigrid (AMG) method should be applied to the original image. The data of the multilayered coarse level are extracted to reconstruct the image with AMG method. Then, background subtraction method is utilized to detect moving targets to eliminate the interference of moving objects. If multiple images appear in the splicing process, the above mentioned procedure is employed to extract the moving targets, and to analyze the trajectory of moving objects. Experimental results show that the method in this paper has strong robustness, and it can effectively remove interference of noise and detect moving targets accurately.