Search

In this paper, we discuss the use of radar sensor to localize an airborne platform using a proposed rotation-invariant template matching algorithm based on data fusion. If a digital elevation model (DEM) under the platform is measured by radar, it can be used to estimate the position of the platform through comparing a pre-recorded DEM of large area through template matching. We implemented four different rotation-invariant template-matching algorithms and compared their performances from the perspective of position estimation accuracy and computational complexity. Based on the investigations, we chose a Fourier transform-based method that utilizes the characteristic of magnitude invariance with the shift of the original data. To improve the position estimation by providing more information, the gradients of DEM were derived. To combine the total error cost from the original DEM and its gradient, we employ an adaptive weighting inspired by the information theory. A higher weight is given to data that have more information. The information is measured as entropy, which is the inverse of the probability that the error cost is the minimum out of all possible error costs in the template-matching process. Through the suggested adaptive weighting, the position estimation accuracy has been increased by 16.04% compared with the uniform weighting. Compared with only using a DEM, the use of gradients of DEM improves the accuracy by 55.37%.

We present a novel affine projection algorithm (APA) which dynamically selects input vectors in order to improve convergence performance. The optimum selection of the input vectors is derived by the largest decrease of the mean-square deviation. The experimental results show that the proposed algorithm has a fast convergence speed and a small steady-state error compared to the conventional APA. In addition, the proposed algorithm retains a low overall computational complexity.

We present a novel affine projection algorithm which reduces complexity by selecting a subset of input regressors at every iteration. The optimal selection of input regressors is derived by comparing the cost functions based on the principle of minimum disturbance. The proposed algorithm shows good convergence performance with various experimental results