摘要
以认知神经科学的研究成果为依据,对视网膜会强烈响应大反差视觉刺激的现象和初级视觉皮层上视觉信息的生成机制进行了模拟,提出了一种基于视觉反差的显著图生成与目标检测方法。本方法可以在不考虑目标的形状、边缘或其他形式先验知识的情况下检测出场景中具有显著反差的目标。在对地场景中车辆目标的检测实验里,将其与另外两种典型的显著图生成方法进行了对比,结果显示,这种基于视觉反差的目标检测方法能够较好地将注意力集中在场景中具有较大反差的目标之上,且具有高效性和鲁棒性。
The phenomenon that the retina will strongly respond to large contrast visual stim- ulation and the generation mechanism of visual information in the primary visual cortex can be simulated. Therefore, In this paper, we propose a method generating saliency map and detecting object according to the cognitive neuroscience research. The method can detect sig- nificant contrast object without considering the shape, edge, or other forms of prior knowl- edge of the objects. We compare the method with the other two typical saliency methods in detecting vehicle target from ground scenes. The results show the detection method based on visual contrast can better concentrate on the object with greater contrast and is fast and ro- bust.
出处
《武汉大学学报(信息科学版)》
EI
CSCD
北大核心
2012年第4期379-383,376,共5页
Geomatics and Information Science of Wuhan University
基金
国家科技支撑计划资助项目(2012BAH35B03)
国家863计划资助项目(2011AA010500)
武汉大学测绘遥感信息工程国家重点实验室开放研究基金资助项目(09重点项目)
武汉大学中央高校基本科研业务费专项资金资助项目(114041)
关键词
目标检测
显著图
视觉反差
视网膜
初级视觉皮层
object detection
saliency map
visual contrast
retina
primary visual cortex