典型的图像显著性检测算法分析和比较

The Analysis and Comparison for Detection Algorithms of Typical Image Saliency

图像的显著性检测是计算机视觉领域最为活跃的研究方向之一,如何检测出复杂背景下的显著物体,同时降低时间复杂度,得到分辨率高、边界清晰、整体均匀突出的显著物体是研究重点。首先论述了显著性检测原理,然后对近年来基于图像的显著性检测研究进行了详细的分析和介绍,将有关方法分成时域和频域两类,并重点介绍了各类中具有高影响力的研究成果,通过在常用的图像显著性检测的数据库中实验对比,分析了各类方法中典型技术的优缺点,最后展望了图像显著性检测在未来的发展方向和趋势。

The detection of image saliency is one of the most active research directions in the field of computer vision. The focus of the study is how to detect salient objects under complex background, and get the high -resolution, clear boundary and the overall symmetrical salient objects while reducing the complexity of time. Firstly, the principle of salient detection is discussed. Then, this paper analyzes and introduces the research of salient detection in recent years in detail. The related research methods are classified into spatial domain and frequency domain, among which the most influenced research re- suits are paid attention to introduction. Through comparing the experiments of common detection of im- age saliency in database, the merit and demerit of typical technology of different ways are found. Final- ly, the developing direction and trends of detection of image saliency in the future is introduced.