基于背景统计的ROI提取方法

A Method for ROI Extraction Based on Background Statistics

针对基于视觉特征的ROI提取算法的不足和图像噪声的影响,借鉴图像分块的思想,提出了一种基于背景像素点统计模型进行ROI提取的算法。该算法基于图像分块,消除了噪声作为显著视觉特征时对ROI提取的影响。将该算法运用于室内监控图像中的ROI提取实验。结果表明,所提算法能实现多种情况下感兴趣区域的自动准确提取。

For the lack of algorithm for ROI extraction based on visual features and for the effect ofimage noises, this paper presents a new algorithm for ROI extraction based on the statistical model of background pixels, drawing on the idea of image block. The algorithm based on image segmentationeliminates the effect of noises, when they become significant visual features, on the ROI extraction.Finally, the algorithm is used in indoor monitoring images for ROI extraction test, and results showthat the algorithm can automatically and accurately extract the region of interest in a variety of circumstances.