Abstract:

Based on the characters of static scenes and block processing techniques, a video segmentation algorithm based on color information and sampling detection was proposed. In the algorithm, every frame was divided into many small regular blocks(Information Detecting Window,IDW) and showed by matrix. According to the detecting results with sampling IDWs, the algorithm quickly rebuild background model and segment moving object with the RGB color and grey scale information. It is shown that the algorithm combines the advantages of less runtime,stronger adaptability and accuracy with comparison to other background subtraction algorithms.

Key words: adaptive background model, information detection window, moving object segmentation, color information

摘要：

提出了一种基于彩色信息和抽样检测的视频分割算法。该算法把每一帧视频图像划分 为M×N个互不重叠的像素块(信息检测窗),并用矩阵表示,对M×N个信息检测窗进行抽样检测, 针对检测结果,快速重建背景模型,依据彩色信息和灰度信息分割运动目标。实验结果表明,该算法 与已有算法相比,具有分割速度快、适应性强、运动目标分割准确等特点。

关键词: 自适应背景模型, 信息检测窗, 运动目标分割, 彩色信息