Abstract. The tracking of multiple maneuvering targets in a dense clutter environment is investigated. An effective parallel processing algorithm based on state fusion and fast joint probabilistic data association (FJPDA) is proposed. State fusion and feedback of all state information are used to fit different movements of targets. The FJPDA, combining cluster matrix decomposition with a fast data association algorithm, is used for tracking multiple targets. The advantages of this algorithm are not only keeping the accurate estimation and fast response for target maneuvering, but also reducing the computational burden of data association from N! to N/4* (4!). Three examples are simulated to prove the validity and reliability of the proposed new algorithm.

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.

In remote sensing community, IHS (intensity, hue, and saturation) transform is one of the most commonly used fusion algorithm. A study on IHS fusion indicates that the color distortion cannot be avoided. Meanwhile, wavelet decomposition has a property of frequency division in transform domain. And the statistical property of wavelet coefficient reflects those significant features. So, a united optimal fusion method, which using the statistical property of wavelet decomposition and IHS transform on pixel and feature levels, is proposed. That is, the high frequency of intensity component I is fused on feature level with multi-resolution wavelet in IHS space, and the low frequency of intensity component I is fused on pixel level with optimal weight coefficients. Spectral information and spatial resolution are two performance indexes of optimal weight coefficients. Experiment results show that it is a practical and effective method.

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.