基于视觉注意的显著物体检测实验设计

The Design of Salient Object Detection Experiment Based on Visual Attention

从图像中快速准确定位出图像主要内容的显著物体检测已经成为数字图像处理的一个研究热点。研究并设计了基于视觉注意的显著物体检测实验,通过实验操作增强学生对显著物体检测的理解以及激发学生进一步学习和研究的兴趣。首先介绍了基于视觉注意的显著性物体检测原理和方法,然后利用视觉注意方法进行显著性检测,在此基础上利用显著物体密度最大化这一特征进行有效子窗口搜索以实现显著物体检测,最后介绍了查准率、查全率和F度量作为性能评价准则并对实验结果进行了分析。提供可实现上述方法的Matlab程序代码,学生可根据相关程序进行显著物体检测处理,得到符合人眼视觉感知的检测结果。该实验方案可以检测不同内容、主题与背景环境的自然图像,具有较强的鲁棒性和较好的可操作性。

It has become an active research focus in the field of image processing to quickly and precisely locate the main information of the image. This paper studied and designed a salient object detection experiment based on the visual attention,which is aimed to deepen students＇ understanding and stimulate their interest in further study on salient object detection. Firstly,the basic principle and implementation of salient object detection based on visual attention are introduced. Moreover,saliency map is produced by visual attention models and then the salient object is carried out by efficient subwindow search algorithm on the basis of the maximization saliency density of saliency map. And finally,the performance is analyzed with the criterion of precision ratio,recall ratio and F-measure. This experiment also provides the Matlab code to help the students achieve results which are in accord with human visual perception. Experimental results show that the proposed method could achieve better performance without regard to image ＇s different content,subject and background,and has better robustness and maneuverability.