Estimation of Displacement Vector Field from Noisy Data using Maximum Likelihood Estimator

The present study proposes an approach for robust motion estimation between two successive image frames, from a degraded sequence. The method is based on Generalized Cross-Correlation (GCC) Methods, where the phase of the Fourier components is used for motion parameter estimation. This method uses "Whitening" FIR filters to sharpen the cross-correlation maximum, thereby improving the accuracy of identification of the peak. The estimators of interest are the Phase Transform (PHAT), and the Maximum Likelihood (ML) estimators. For robust motion estimation it has been found that the ML estimator is particularly suited to this purpose. The accuracy of the estimators is also discussed. Significant results have been obtained for sub-pixel translation of images of different nature and across different spectral bands.

I. Introduction

In order to build a video coder that is robust in the presence of noise, the motion estimation process must be able to track objects within a noisy source. In a noisy source, objects appear to change from frame to frame because of the noise, not necessarily as the result of object motion [1]. Noise gets added to video in the process of recording it. This problem is even more acute when converting from video on analog tapes to video in digital format. Noise is undesirable not only because it degrades the visual quality of the video but also because it degrades the performance of subsequent processing such as compression [2]. Many motion estimation schemes have been developed. They can be classified into spatial-domain and frequency-domain approaches. The spatial domain algorithms consist of matching algorithms and gradient-based algorithms. The frequency domain algorithms consist of phase correlation algorithms, wavelet transform-based algorithms, and DCT-based algorithms [3].