| In order to drive along the airport border safely, infrared camera and visible camera on the patrol are selected as platform for the study. Study applications about compressed sensing in dual-source image fusion and propose dual-source image fusion algorithm based on redundant dictionary in road detection.Firstly, study some basic questions such as coordinate system definitions of imaging model, the linear camera model and distortion model. Use Zhang Zhengyou calibration method based on planar templates for experiments, demarcate video camera on patrol car and get the camera parameters.Then build road model. B-spline model is used to build the road model of airport border due to the local control and flexibility. Then choose different road detection algorithm depending on the characteristics of infrared images and visible images, get the edge of the road and fit with the B-spline model established before. Finally, get the road detection inadequacies of infrared and visible images by experiments on different seasons, different structures.Dual-source image fusion method based on redundant dictionary is proposed for the lack of infrared and visible images detection algorithm. First, construct redundant dictionary, select the appropriate wavelet sparse base and sparse infrared visible light images. Orthogonal matching pursuit algorithm is selected to reconstruct after fusion and verify the effect of the fused image by road edge detection finally. Select a great number of infrared and visible images under different lighting conditions for experiment. Experimental results show that the dual-source image fusion algorithm based on redundant dictionary can fit for a variety of lighting conditions, has clear image details and can detect road borders accurately. |
PDF Full Text Request