MitchellNetravali

Mitchell–Netravali reconstruction filter is a two-parameter family of cubic convolution kernels used for resampling digital images and textures in computer graphics and image processing. Introduced by Mitchell and Netravali in the late 1980s, it is designed to balance sharpness and fidelity while reducing artifacts such as ringing and overshoot that can occur during resampling.

The filter is defined by two parameters, B (smoothing) and C (ringing control). It produces a piecewise cubic kernel with support over four pixels, enabling smooth interpolation while preserving edge detail. The most commonly cited settings are B = 1/3 and C = 1/3, which yield a well-balanced trade-off between sharpness and artifact suppression. Other parameter choices can tune the trade-off toward crisper detail or smoother results, depending on the application.

Applications of the Mitchell–Netravali filter include image upsampling and downsampling, texture filtering in computer graphics, and various stages of digital image processing pipelines. It is valued for offering good visual quality with relatively modest computational requirements, and it has influenced many resampling implementations in software and hardware.

Legacy and context: the filter is one of several cubic convolution kernels used for image reconstruction, alongside approaches such as bicubic, Catmull–Rom, and Lanczos methods. It remains a reference point in discussions of interpolation quality and artifact control. The filter is named for its developers, Mitchell and Netravali, whose collaboration produced a widely adopted solution in image reconstruction and resampling.