Polyphase antialiasing in resampling of images

Area of Science:

• Digital Image Processing
• Signal Processing

Background:

• Image resizing commonly uses small interpolation kernels, leading to poor frequency domain characteristics.
• Finite-length interpolation kernels cause aliasing, resulting in jagged edges and sampling noise during image enlargement.
• Aliasing degrades image quality, particularly when resizing by a rational factor (L/M).

Purpose of the Study:

• To reduce aliasing effects in image resizing.
• To improve image quality without requiring large interpolation kernels.
• To offer a simpler approach to aliasing reduction in digital image processing.

Main Methods:

• Utilizing a polyphase representation of the interpolation process.
• Separately treating the polyphase filters for aliasing reduction.
• Analyzing the one-dimensional case for separable interpolation.

Main Results:

• A considerable reduction in aliasing effects was achieved.
• The method is effective even with small interpolation kernel sizes.
• The proposed procedure is simple to implement.

Conclusions:

• The polyphase approach effectively mitigates aliasing in image resizing.
• This method offers a practical solution for enhancing image quality.
• The findings are applicable to separable interpolation and one-dimensional cases.