contraststretching

Contrast stretching, or normalization, is an image processing operation that increases image contrast by expanding the range of displayed intensities. It redistributes pixel values so that the available dynamic range is used more fully, making details easier to see. The technique is commonly applied to grayscale images and can also be used on color images by adjusting luminance or each channel.

Linear contrast stretching maps the image minimum to the darkest display level and the maximum to the brightest. For an 8-bit image, the typical formula is y = (x - min) × 255 / (max - min). In practice, min and max can be chosen from percentiles to reduce the influence of outliers.

Piecewise linear stretching uses multiple segments to control the slope in different intensity ranges, producing an S-shaped response that can preserve both shadows and highlights. Other variants include sigmoid and histogram-based stretching, which adjust values according to their distribution. In medical imaging, a related technique called windowing maps a chosen intensity range to the display scale.

Application domains include photography, satellite imagery, and medical imaging, where improved contrast aids interpretation. For color images, stretching is typically applied to the luminance channel or to each channel separately with care to minimize color distortion.

Limitations include amplification of noise, distortion of absolute intensities, and dependence on chosen endpoints or percentiles. It is most effective when the input image contains meaningful extremes that should be displayed across the full range.