Infrared/Passlve Millimeter Wave Image Fusion Based On Frame Multi-resolution Analysis

This paper mainly considers IR/PMMW image fusion algorithm, it uses the IR and PMMW detection tank image to make experiment, its research contents as follows:Combine the principal component analysis transform with the multi-resolution analysis, this paper puts forward a new MRA-based PCA (M-P) fusion method. First, it de-noises the IR and PMMW detection tank group images. Second, transforms the reconstructed images with PCA respectively. Third, replaces the first principal component of low resolution image for that of high resolution image. Finally, the fusion image is obtained by taking the inverse PCA transform. The experiment indicates that compared with traditional PCA method, the new fusion method can better distinguish the tank group targets from the surface features.Laplacian pyramid transform is applied for the fusion of IR image and PMMW image. First, this paper makes Laplacian pyramid decomposition of original images. Then, the top decomposed image of IR image is used as the top image of the fusion result and other layer images are fused by selecting their pixel with maximum absolute value on IR and PMMW image corresponding to the layer. After that, the fusion image is obtained by taking the inverse Laplacian pyramid transform. The experiment indicates that the fusion result can better distinguish the targets from the features.Discrete wavelet frame transform is applied for the fusion of IR image and PMMW image. First, this paper de-noises the IR and PMMW images using wiener filtering, respectively. Second, decomposes the de-noised images with discrete wavelet frame transform. Then, the low-frequency components of the decomposed IR image is used as that of the fusion result and high-frequency components are fused by selecting their pixel with maximum absolute value. After that, the fusion image is obtained by taking the inverse discrete wavelet frame transform. The experiment indicates that the fusion result of discrete wavelet frame transform-based has better performance than that of Laplacian pyramid transform-based both in subjective visual effect and objective evaluation criterion.This paper puts forward a lifting wavelet transform-based fusion method to fuse the PMMW image and visible image for concealed contraband detection used in security check. First, this paper makes lifting wavelet transform decomposition to the original images. Second, the corresponding fusion rules based on pixel with maximum absolute value and region variance are established for the fusion of low-frequency and high-frequency components, respectively. Third, the fusion image is obtained by taking the inverse lifting wavelet transform. Finally, this paper enhances the contrast of the fused image. The experiment indicates that the method can integrate the clear visual effect of visible image and the significant target outline of the PMMW image into the fused image, it has realized speedy identification of the concealed.