A new image registration method robust to noise

Abstract

-Image registration is a fundamental task in many image processing applications. In this paper, we estimate the translation, rotation, and scaling parameters between a reference image and a distorted image in order to register them. We use the ratio of means to estimate the scaling factor, the Radon transform to estimate the rotation angle, and the fast Fourier transform (FFT) to estimate global spatial shifts. Experimental results show that our proposed method can estimate the transformation parameters very accurately. Furthermore, our method performs very well in noisy environments.

Similarity transformation parameters recovery based on Radon transform. Application in image registration and object recognition

“…The estimation results are then employed in image registration. As it is reported in Table 3, it appears that the results given for both images outperform the results obtained in [7] for the scaling factor α and the rotation angle ϕ 0 with or without noise. The outstanding performance of the proposed method is also reported for the translation vector estimate in case of Lena image.…”

“…In fact, the robustness of RT to additive noise argued in [18,19] can have a direct effect on the robustness of our algorithm to this type of noise since it uses exclusively the Radon projections. In another experiment, the performance of the proposed algorithm is compared with the performance of the method proposed in [7,21] on the recovery of RST parameters on Barbara and Lena images of size 512 Â 512, which are subject to a sequence of RST transformations, as illustrated in Fig. 9.…”

“…The outstanding performance of the proposed method is also reported for the translation vector estimate in case of Lena image. Considering in [7] that the estimates of the translation vector components are rounded to integer values, their inherent errors could be bigger compared to those of the proposed algorithm, for which the maximal errors on the x or y coordinates and the translation vector norm are less than 0.45 and 0.24 (in pixel), respectively. In addition, we compare our method with the work in [21] which deals with the design of an RST-invariant shape descriptor from the combination the Radon transform, the amplitude extraction and the log-mapping.…”

“…Even if the computational time of the proposed algorithm remains weak enough as shown in the experiments, it may be further reduced. In fact, the running Method in [7] Method in [21] Proposed method…”

“…The techniques designed to recover spatial transformations linking two patterns or two images using directly their Radon transforms have received recently a lot of interest and are increasingly investigated. For image registration tasks [6], the authors in [7] estimate the RST parameters between a reference image and a distorted image. Here, only the rotation angle is estimated using Radon transform.…”