By fusing the visible and infrared images to improve the detection and recognition ability of equipments is a focus research. Aiming at the contradictory aspects of the current fusion algorithm for speed and accuracy, a fusion algorithm of visible and infrared images is proposed. Firstly, the visible and infrared imaging systems’ calibration information is obtained by calibrating the system. Secondly, the correspondence between visible image and infrared image pixel is established through constructing mathematical model. Finally, on the basis of Laplace decomposition, the visible and infrared images is fused with regional energy optimization principle. The experimental result shows that the proposed method increases the speed by 22.9%, improving the real-time performance of the fusion algorithms, while remains registration accuracy unchanged.
A dimensionality reduction method is proposed by using the second generation Bandelet transform. The redundant components of the hyperspectral cube are firstly partitioned into several subsets. Subsequently the Bandelet coefficients and the geometries flows of the hyperspectral image are generated by performing second generation Bandelet transform. In the follow step, Principal Components Analysis (PCA) is introduced to simplify the redundant data. Finally, the new reduced hyperspectral cube is reconstructed by taking inverse Bandelet transform. Some numerical simulations are made to test the validity and capability of the proposed dimensionality reduction algorithm.
Vanishing point is an important concept in the sensors’ self-calibration method. For traditional self-calibration method
based on vanishing point, it's computationally intensive, real-time not high and noise sensitive. A novel method of sensor
self-calibration based on rectangular vanishing point characteristics is proposed to solve all these problems mentioned
above. Firstly, four template images are captured from different angles and locations. Subsequently, the vanishing points
are calculated by using the coordinates of the four vertices in rectangle. Finally, the novel sensor parameter equation is
proposed by using the geometry properties of rectangle and the vanishing points. Some numerical simulations are made
to test the validity and robustness of the proposed algorithm.
A fusion algorithm of hyperspectral and high-resolution images based on principal component analysis (PCA) and second generation Bandelet transform is proposed. Primarily, the numerous components of the hyperspectral image are divided. Subsequently, the maximum rule is used to select the Bandelet coefficients and geometry flows of the hyperspectral image which are transformed by PCA in the following step. Finally, the fused image is reconstructed by taking inverse PCA and Bandelet transform. Some numerical simulations are made to test the validity and capability of the proposed fusion algorithm.