A novel image fusion method based on an expectation maximization (EM) algorithm and the discrete wavelet frame (DWF) transform is proposed. The registered images are first decomposed using the DWF transform, which is both aliasing-free and translation-invariant. The DWF decomposes the image signal into a multiresolution representation with both low-frequency coarse information and high-frequency detail information. The EM algorithm is used to fuse the low-frequency coarse information of the registered images. The informative importance measure is applied to fuse the high-frequency detail information of the registered images. The final fused image is obtained by taking the inverse transform of the composite multiresolution representations. Simulation results show that the proposed method outperforms the conventional image fusion methods.

Fusion of synthetic aperture radar (SAR) and forward looking infrared (FLIR) images is an important subject for aerospace and sensor surveillance. This paper presents a scheme to achieve a natural color image based on the contours feature of SAR and the target region feature of FLIR so that the overall scene recognition and situational awareness can be improved. The SAR and FLIR images are first decomposed into steerable pyramids, and the contour maps in the SAR image and the region maps in the FLIR image are calculated. The contour and region features are fused at each level of the steerable pyramids. A color image is then formed by transferring daytime color to the monochromic image by using the natural color transfer technique. Experimental results show that the proposed method is effective in providing a color fusion of SAR and FLIR images.

In this paper a fusion method is proposed for merging a high-resolution panchromatic image and a low- resolution multispectral image. The algorithm is based on discrete wavelet transform (DWT). It uses correlation moment rule to the low frequency bands and local deviation rule to the high frequency bands separately Enperimental results indicate that the proposed approach outperforms the traditional methods.

A new fast high-order neural network learning algorithm for pattern recognition is proposed. The new learning algorithm uses some properties of trigonometry for reducing and controlling the number of weights of a third-order network used for invariant pattern recognition. Experimental results on typed upper case English letters indicate that the new approach maintains the higher classification accuracy and reduces the complexity of neural networks significantly. The proposed method can also be adapted for applications in some other pattern recognition problems.

An adaptive algorithm for tracking maneuvering targets is proposed. This algorithm is implemented with two filters and a multilayer feedforward neural network using state fusion, together with the current statistic model and adaptive filtering. The neural network fuses automatically all the state information of the two filters and tunes adaptively the system variance for one of the two filters to adapt to different target maneuvers when the two filters track the same maneuvering target in parallel. Simulation results show that the adaptive algorithm tracks very well maneuvering targets over a wide range of maneuvers with high precision, in both one and three-dimensional cases.