Image resizing is the downscaling or upscaling of an image. Dozens of algorithms exist, all with performance vs. quality tradeoffs.
Image resizing (also called image scaling) is the downsampling or up-sampling of an image using an interpolation algorithm. Common downscaling algorithms are Lanczos, Bicubic Sharper, Bicubic Smoother, Fant, Bicubic, Bilinear, and NearestNeighbor (listed in order of average result quality). Fractal algorithms often produce better results for upscaling photos, while there are many algorithms optimized for upscaling pixel art.
Due to algorithmic complexity, most developers use library implementations of these algorithms.
FreeImage
Offers CatmullRom, Lanczos3, bspline, box, bicubic, and bilinear filters. FreeImage focuses on implementation simplicity, and doesn't use hardware accelerations or SIMD extensions, so these speeds may not be acceptable for real-time display of images.
Windows GDI+
While loathed by many, GDI+ does include an excellent 2-pass Bicubic filter. It offers: 2-pass Bicubic, 1-pass bicubic, bilinear, and nearest neighbor. Those using it should be aware that it may add rings to images unless WrapMode is set to TileXY, as otherwise the algorithm samples data from outside the image bounds.
WIC (Windows Imaging Components)
The IWICBitmapScaler class offers Fant, 1-pass bicubic, bilinear, and nearest neighbor. Implementations are optimized for performance, and can be 2-4x faster than their GDI counterparts, although there isn't currently a 2-pass bicubic filter such as offered by GDI+
