Optical EngineeringAlias-free image subsampling using Fourier-based windowing methods
|Format||Member Price||Non-Member Price|
An antialiasing image subsampling method using Fourier-based windowing is presented. To reduce resolution, directly subsampling an image results in aliased data. To avoid aliasing, the original image has to be low-pass filtered within the bandwidth of the subsampled image prior to subsampling. However, applying a simple rectangular filter causes ringing effect (ripples), which is also called the Gibbs phenomenon. Conventional finite impulse response (FIR) or infinite impulse response (IIR) filters use spectral windows (e.g., Gaussian, Hamming, Hanning, or Butterworth windows, etc.) to yield a smooth transition band at the cutoff frequency. Yet this approach attenuates the higher frequency contents of the subsampled image and results in the widening of the transition band of the impulse response. An alias-free image subsampling method is developed using a power window to reduce the spatial domain ringing effect and to preserve most of the spatial frequency contents of the subsampled image.