结合超像素和直方图阈值的显著区域检测算法

Superpixel and histogram threshold based salient region detection

由于现有显著性检测算法得到的显著图内容差异较大,因此设计一种具有普遍适用性的显著区域检测算法以依据不同稀疏度的显著图进行高效率的检测仍是一个具有挑战性的问题。提出结合超像素分割方法和直方图阈值化分割方法以在不同的显著图上进行显著区域检测并提高检测效率。利用超像素分割方法对原图像进行分割计算,计算每个超像素的平均显著度值,并用该平均值取代超像素内每个像素的原像素值更新显著图,利用新显著图的直方图将显著图二值化以确定显著目标,利用一覆盖显著目标的最小矩形区域表示检测得到的显著区域。实验结果表明,在不同的显著图上,所提算法能有效检测显著区域,在检测效果的客观度量指标和时间性能指标上均优于现有算法。

The saliency maps achieved by different saliency algorithms are obviously different, so a salient region detec-tion algorithm with good performance which can be applied to different saliency maps is necessary. This paper proposes a novel algorithm based upon the superpixel segmentation and histogram threshold to enhance the algorithm efficiency. The proposed algorithm uses superpixel algorithm for segmenting image, computes each pixel’s average saliency of each seg-mented region according to the saliency map, then utilizes the histogram technique to obtain the binary map which indi-cates the salient object, and obtains the salient region using a minimum rectangular window which covers the salient object. Experimental results show that compared to the current schemes the presented approach can detect the salient region in a better way and has a good performance of precision, recall, F-measure and computational efficiency.