目前的显著性检测算法主要依赖像素间的相互对比,缺乏对显著目标自身特性的分析理解.依据显著目标是显眼、紧凑和完整的思路,提出一种基于目标全局孤立性和局部同质性的随机游走显著目标检测算法,将视觉显著性检测公式化为马尔科夫随机游走问题.首先将输入图像进行分块,根据像素块之间颜色特征和方向特征的相似性确定边的权重,从而构建图模型；然后通过仝连通图搜索提取全局特性,突岀全局较孤立的区域；同时通过k-regular图搜索提取局部特性,增强局部较均匀的区域；最后将全局特性和局部特性相结合得到显著图,进而确定感兴趣区域位置.实验结果表明,相比于其他两种具有代表性的算法,所提方法检测结果更加准确、合理,证明该算法切实可行.%The existing saliency detection algorithm mainly focuses on the inter-pixel contrast and lacks global perspective for analyzing and understanding the object in complex surroundings. According to the thought that a salient object in an image is often conspicuous and compact, an unsupervised graph presentation random walk salient object extraction algorithm based on global isolation and local homogeneity is proposed, and the problem of salient region detection is formulated as Markov random walk. First of all, the graph model is formed by dividing the input image into block images and using color and orientation features to determine the weight of edge, and then the isolated regions are obtained by using the random walk on a complete graph to extract the global properties of the image. Meanwhile, the uniform regions are enhanced by using the random walk on a k-regular graph to extract the local properties of the image. Finally, the saliency map is obtained by combining the global properties and local properties of the image, and the salient object is located and extracted according to the saliency map. Experimental results show that the proposed algorithm is more reasonable and effective than the two representative methods for salient object detection.
展开▼
