Abstract: With the development of computer vision, image salient region detection plays an important role in the field of image processing. In order to detect the salient region of natural image exactly, a method of saliency detection based on region contrast is proposed. The method first processes the input image using superpixel segmentation, then calculates the region contrast by features of color and spatial, then combined with the distance of average vectors of pixel features between image sub-region and its neighborhood and the center prior to generate high quality saliency map. Compared with the other saliency detection methods, the experimental results show that this method can detect the salient region more sufficient and better suppress the background.

Key words: saliency detection, region contrast, center prior, saliency map

摘要： 随着计算机视觉的发展，图像显著区域检测在图像处理领域越来越重要。为了对自然图像中的显著区域进行准确的检测，提出了一种基于区域对比的图像显著性检测方法。首先对图像进行超像素分割预处理，然后利用图像的颜色特征和空间特征算出区域对比度，再结合图像子区域与其邻域像素平均特征向量的距离以及中心优先原则得到图像高质量的显著图。仿真实验结果表明，与其他的显著性检测算法相比，可以更加有效地检测出显著性目标，更好地抑制背景。

关键词: 显著性检测, 区域对比, 中心优先, 显著图