Image resizing is a fundamental operation in digital imaging that involves changing the pixel dimensions of an image, and the quality of the result depends entirely on the resampling algorithm used. Resampling is necessary because digital images are discrete grids of pixels, and changing dimensions requires calculating new pixel values that did not exist in the original data. The mathematics behind this process is called interpolation, and different algorithms produce dramatically different results in terms of sharpness, artifact prevention, and processing speed.

Bilinear interpolation is one of the simplest high-quality resampling methods. For each pixel in the output image, it maps back to a position in the source image that typically falls between four source pixels. It then computes a weighted average of these four neighbors based on distance, producing smooth results. While fast and adequate for many uses, bilinear interpolation can produce slightly soft results because it effectively applies a mild blur during the averaging process.

Bicubic interpolation improves on bilinear by considering a 4x4 neighborhood of 16 source pixels instead of just 4. It uses cubic polynomial functions to compute weights, producing sharper results with better preservation of edges and fine detail. Bicubic interpolation is the default in most professional image editing software because it offers an excellent balance between quality and computational cost. The cubic weighting function can be tuned with a parameter (often called alpha or a) that controls the sharpness-versus-ringing tradeoff.

Lanczos resampling represents the highest quality commonly available. It uses a sinc function (sin(x)/x) windowed by a Lanczos window, typically considering a 6x6 or 8x8 neighborhood of source pixels. The sinc function is mathematically the ideal interpolation kernel for band-limited signals, making Lanczos theoretically optimal for preserving detail. However, it can introduce slight ringing artifacts (halos around high-contrast edges) and is computationally more expensive. The distinction between resolution and dimensions is also important to understand. Resolution refers to pixel density (pixels per inch or PPI), while dimensions refer to the total pixel count (width times height). An image can have large dimensions but low resolution if intended for screen display, or small dimensions but high resolution for print. Resizing changes dimensions, and the effective resolution depends on the output medium.