Automatic Infrared Condensing Tower Target Recognition using Gradient Vector Features

Condensing towers are such kind of special buildings whose shape can be well described using standard parameter equation according to corresponding architectural specifications.In order to perform automatic condensing tower target recognition,automatic target acquirement method based on template matching is proposed,that is to match real-time infrared images with visible satellite images.The template used here is generated only from condensing tower's parameter description.A gradient vector correlation method is also proposed to perform similarity measurement between infrared and visible images,which played an important role in automatic infrared targets acquirement.The proposed method takes gradient images of the multi-sensor images as common features.To avoid losing direction information,the image gradient field is considered as a two-dimensional vector matrix,and the gradient image's value and direction information are both used to computer correlation coefficient.Experimental results show that the proposed method achieves good performances for both day and night infrared images.