Camouflage texture evaluation using saliency map
Proceedings of the Fifth International Conference on Internet Multimedia Computing and Service
| Hi-index | 0.00 |
This paper proposes a vision model for object detection based on scale-space theory, considering knowledge obtained in neurophysiology and human visual characteristics obtained in visual psychology. The proposed model is principally composed of (a) the early vision model and (b) the attention calculation model. In this paper, it is shown first that the visual characteristics can be described by a discretized scale space, considering their multichannel property, spatial nonuniformity, and orientation selectivity. The early vision model is formulated. Next, the attention calculation model and the operation algorithm are formulated on the basis of the knowledge obtained by scale-space theory. The numerical experiments reveal that the proposed model has the following properties. (i) For objects with a high intensity difference, the center of overt attention moves to the center of the object. (ii) The spatial extent of overt/covert attention can be calculated adequately. (iii) The object is captured at the central region of the retina, where the resolution is the highest. Since the proposed model is based on scale-space theory, the theory or model can be easily extended. There are other advantages from an engineering standpoint, such as simple structure, easy implementation, small computational requirements, and very few parameters to be adjusted. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(10): 1–13, 2004; Published online in Wiley InterScience (). DOI 10.1002/scj.10708