基于颜色和空间距离的显著性区域固定阈值分割算法

Color and Space Distance Based Salient Region Detection Using Fixed Threshold Segmentation

针对现有的基于空间域的显著性检测算法在分割显著性区域时需要依赖图像分割算法的不足,提出一种基于颜色和空间距离的显著性区域固定阈值分割算法。该算法首先对图像建立图像金字塔,并对每层的图像进行颜色量化和图像分块的预处理;然后利用颜色和空间距离计算得到显著性图;最后进行阈值分割,得到显著性区域。在MSRA1000公开数据集上的实验结果表明,该算法在精度、召回率和F测度方面的表现均优于现有的几种算法。因此,提出的算法在检测效果上优于现有的显著性区域检测算法,而且可以简单地分割出显著性区域。

Against to the problem that the existing salient region detection algorithm based on spatial domain depends on image segmentation algorithm in segmentation of salient region,we proposed an algorithm based on color and space distance for detecting salient region using the fixed threshold segmentation algorithm.By this algorithm,spatial scales are created using dyadic Gaussian pyramids and the color is quantized and image is blocked at each scale.Then,the color and space distance of image patches are calculated in the each scale.Finally the salient region is segmented.Experimental results in the MSRA1000 public database show that the performance of the algorithm is better than others in precision,recall and F-measure.Therefore,the proposed algorithm is comparable to the state-of-art salient region detection algorithms and can segment the salient region simply.