An Attention Target Detection Method Based on Dynamic Saliency Map

Abstract:

Thinking of the characteristics of the human visual system, proposed a target detection model of attention mechanism which was based on dynamic saliency map. This method improved classical visual attention calculation model, extracted the static characteristics of intensity, color and orientation, and selected different parameters to fuse into a static saliency map which was based on different target on scenarios. Using differential filter method to extract the dynamic features of two images, and fused different scales feature maps into the dynamic saliency map. At the end, with modified factor modify two saliency map, and fused into the basic image with which detected the moving targets. This method simulated human’s attention mechanism, extracted different scales features with strong processing and analysis capacity. Experimental results show that the method can quickly and accurately detect moving target, can effectively meet the single target detection and multi-target detection.

• Periodical: Advanced Materials Research (Volumes 308-310)
• Pages: 574-578
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.308-310.574
• Online since: August 2011
• Authors: Hong Chang Ke, Hui Wang, Hong Yu Li
• Keywords: Attention Mechanism, Dynamic Saliency Map, Salient Features, Target Detection

References

[1] Juehui Hu, Yimin Li and Xiaolu Pan:Science Technology and Engineering2010, vol.15(2010), p.794

[2] Jiayi Lin, Zhezhou Yu, Jian Zhang and Anna Ma:Chinese Journal of scientific Instrument, vol.29 (2008), p.111

[3] L. Itti and C. Koch: Nature Reviews Neuroscience, 2001, vol.2(2001), p.194

[4] L. Itti, C. Koch and E. Niebur: Transs on Pattern Analysis and Machine Intelligence, vol.20(1998), p.1254

[5] L. Itti and C. Koch: Vision Research, 2000, vol. 40(2000), pp.1489-1506

[6] Peng Zhang and Runsheng Wang: Journals of Software, 2004, vol.15 (2004) : p.891

[7] John K. Tsotsos: (Cognitive Vision Systems, LNCS 3948, 2006).

[8] V. K. Singh, S. Maji and A. Mukerjee: (Third Canadian Conference on Computer and Robot Vision, 2006)