Abstract:

Due to the blank of template extraction technology,most efficient template matching algorithms are con-structed on the basis of artificial template extraction.The defects of current template extraction may reduce theaccuracy and stability of image recognition through their progressively spreading in recognition process,and finallyaffect the results of traffic state identification on the basis of remote sensing images.In order to solve this problem,a remote sensing image recognition method for vehicles is proposed on the basis of self- feedback template extrac-tion,and the correctness of self- feedback template extraction is demonstrated by mathematical derivations.Then,ahigh- resolution remote sensing image recognition and a traffic flow identification are carried out for several certainroad sections on the platform of Matlab.Finally,the feasibility and effectiveness of the proposed method are verifiedthrough analyzing the remote sensing image recognition results of several road sections.

Key words: remote sensing image recognition, self- feedback, template extraction, traffic state identification