基于动态指导滤波的显著性检测方法  

Saliency detection method based on dynamic guided filtering

作　　者：王晨[1,2] 樊养余[1] WANG Chen 1,2 , FAN Yangyu 1(1. School of Electronics and Information, Northwestern Polytechnical University, Xi’an 710072, China;2. School of Aeronautics and Astronautics Engineering, Air Force Engineering University, Xi’an 710038, Chin)

机构地区：[1]西北工业大学电子信息学院,陕西西安710072 [2]空军工程大学航空航天工程学院,陕西西安710038

出　　处：《系统工程与电子技术》2018年第6期1391-1397,共7页Systems Engineering and Electronics

基　　金：国家自然科学基金(61379104)资助课题

摘　　要：显著性检测是指自动提取未知场景中符合人类视觉习惯的兴趣目标的方法。为了进一步提高检测的有效性,同时降低像素类检测算法的计算量和复杂度,提出了基于动态指导滤波的图像显著性检测方法。在新设计的简单迭代指导滤波中,核函数不再像经典指导滤波器那样只利用固定的指导图像,而是利用了输入图像和动态指导图像的联合结构信息,它保证了指导图像对原输入图像较好的结构传递性。其次,为了节约算法的时间成本,采用采样的方式降低算法计算中需要的计算量。最后,为了提取更有效地的显著性区域,引入了关键显著性区域提取方法,通过修正关键点集合得到更准确的目标区域。实验结果表明,相比于其他像素类的显著性检测方法,该算法可以更快速和有效地检测出显著性目标。Saliency detection is to find the most important object automatically according to the human visual habit in unknown scenes.In order to further improve the effectiveness of saliency detection and reduce the computational complexity of the pixel-based detection algorithm,this paper proposes a saliency detection method based on dynamic guided filtering.In the novel simple iterative guided filtering,the kernel function uses the input image and dynamic image to substitute the fixed guided image like the classic guided filtering.This ensures the better structure transmission from the input image to the guided image.Secondly,in order to save the time cost of the algorithm,the sampling method is used to reduce the amount of calculation needed.Finally,in order to extract more effective saliency regions,a key saliency region extraction method is introduced.It can obtain the more accurate object region by modifying the key points set.The experimental results show that the proposed algorithm is more quick and more effective compared with other pixel-based saliency detection methods.

关 键 词：显著性检测 动态指导滤波 指导图像 关键显著性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]