基于多特征的巴文化图像显著性目标检测研究

The research on salient object detection of Ba culture image based on multi-feature

巴文化在川东地区有着重要的研究和应用价值,其图像的显著性目标检测更是数字保护的基础。巴文化图像的背景和前景颜色差别不大,为了快速提取显著性区域,该文提取了多种颜色特征并进行比较,发现局部对比度、局部直方图和颜色分布对巴文化图像较有效。同时该文还提取了边框、纹理两种特征作为颜色的补充,并对3种特征进行了线性融合和贝叶斯融合,将其结果与FT、CA算法进行了比较,发现该文的显著图优于后两者。

The Ba culture has important research and application value in eastern Sichuan,and its image’s salient object detection is the basis of digital protection. There is little difference between the background and foreground colors of Ba cultural images. In order to quickly extract salient areas,this paper extracts a variety of color features and compares them. It is found that local contrast,local histograms and color distribution are more effective for Ba cultural images. At the same time,this paper also extracted two features of border and texture as color supplements,and performed linear fusion and Bayesian fusion of the three features. The results were compared with the FT and CA algorithms,it is found that the saliency map in this paper is better than the latter two.