This paper presents analytic expressions for phase correlation of downsampled images, showing that the signal power in the phase correlation is not concentrated in a single peak but in several coherent peaks. These peaks correspond to the polyphase transform of a filtered unit impulse centered at the registration point. The results provide a closed-form solution for subpixel translation estimation and are used for detailed error analysis. The method is effective for subpixel registration across different spectral bands and image types. The phase correlation method is extended to subpixel registration by deriving analytic expressions for downsampled images. The method is robust to noise and can handle blurred images if the blurring kernel is invariant. The phase correlation method is shown to yield a downsampled version of a filtered unit impulse, leading to multiple coherent peaks. The method is applied to various images, including satellite images across different spectral bands, and experimental results demonstrate its effectiveness. The method is also robust to luminance variations and aliasing errors, which are analyzed in detail. The paper concludes that the phase correlation method provides high accuracy and efficient multispectral registration, although it is limited to translations and requires careful application.This paper presents analytic expressions for phase correlation of downsampled images, showing that the signal power in the phase correlation is not concentrated in a single peak but in several coherent peaks. These peaks correspond to the polyphase transform of a filtered unit impulse centered at the registration point. The results provide a closed-form solution for subpixel translation estimation and are used for detailed error analysis. The method is effective for subpixel registration across different spectral bands and image types. The phase correlation method is extended to subpixel registration by deriving analytic expressions for downsampled images. The method is robust to noise and can handle blurred images if the blurring kernel is invariant. The phase correlation method is shown to yield a downsampled version of a filtered unit impulse, leading to multiple coherent peaks. The method is applied to various images, including satellite images across different spectral bands, and experimental results demonstrate its effectiveness. The method is also robust to luminance variations and aliasing errors, which are analyzed in detail. The paper concludes that the phase correlation method provides high accuracy and efficient multispectral registration, although it is limited to translations and requires careful application.