自动特征选择和加权的图像显著区域检测

视觉显著性度量是图像显著区域提取中的一个关键问题，现有的方法主要根据图像的底层视觉特征，构造相应的显著图。不同的特征对视觉显著性的贡献是不同的，为此提出一种能够自动进行特征选择和加权的图像显著区域检测方法。提取图像的亮度、颜色和方向等特征，构造相应的特征显著图。提出一种新的特征融合策略，动态计算各特征显著图的权值，整合得到最终的显著图，检测出图像中的显著区域。在多幅自然图像上进行实验，实验结果表明，该方法在运算速度和检测效果方面都取得了不错的效果。

Salient region detection with automatic feature selection and weighting

Visual saliency calculation is a key technology in salient region detection.Most of the methods are based on sa-liency map using early visual features.Different features have different contribution to saliency,so a new approach for salient region detection with automatic feature selection and weighting is proposed.Visual features like color,intensity and orientation are extracted.The corresponding feature saliency maps are computed.These feature saliency maps are integrated into a final sa-liency map based a feature integration strategy.According to the final saliency map,the salient regions in an image are detect-ed.This approach has been tested on natural images.The experimental results show that the approach performs well in speed and detection results.