图像感兴趣区域自动提取算法

Automatic Extraction of Regions of Interest

感兴趣区域提取技术在图像处理和分析领域中有着重要地位。提出了一种ROI的自动提取算法,在显著度图和相对位置指示图的基础上,采用基于子区域的区域生长法,以像素点的显著度、相对位置及颜色纹理信息作为生长条件,颜色、纹理信息的权重可以根据图像的内容自适应调整。实验结果表明,该方法与现有算法相比,在速度和ROI提取的准确性方面均有提高。

The extraction technique of ROI plays an important role in image analysis. An automatic extraction of ROI algorithm is proposed. Based on the saliency map and relative position map, the subregion based region grow method is used, besided the saliency map and relative positions, joined the information of color and texture. The power of color and texture can be adjusted by image content automatically. The exoeriment Droves that this altorithm is efficient and effective.