| The performance of visual-target search and localization under complex scene is affected by both the properties of the target itself and its surrounding visual environment. The non-salience of a target is the main reason for the low performance of its identification. The visual attention selection mechanism can lead the attention to a salient object firstly, and the search efficiency of the salient object will be improved. Indirect search provides a method, which uses the relationship between the middle object and the target to search. This paper proposes a progressive search mechanism for target search and localization under complex visual scene, which uses the visual attention selection mechanism and the indirect search method.This method has four main features:(1) A search path is established in searching for objects with high saliency based on saliency analysis;(2) In the process of target search and localization, the quantitative relationship between object locations is utilized to reduce the size of search regions step by step and, simultaneously, enhance the saliency of objects inside regions;(3) Using the location information of target and the location, posture information, their errors of the camera to estimate the original search region of the target, through the estimating, the search region of the target can be narrowed, and the efficiency of target search can be enhanced;(4) Using the GPS data coming from Google Earth software and the image-based modeling techniques to create the3D model of the target scene, and it can provide the localization information of the objects and the location relationships between the objects.Experimental results indicate that the progressive search mechanism we proposes can solve the target search and localization under complex visual scene effectively, and the original search region estimated from the errors of the location and posture is narrowed obvious, and the efficiency of target search is enhanced. |
PDF Full Text Request