I1I2I3 color space

from Wikipedia, the free encyclopedia

The I1I2I3 color space is a computer-optimized space for image processing .

Explanation

Since color spaces that colors by the human perception appropriate parameters (such. As hue , brightness , saturation ) type, for many image-processing systems are unsuitable for this purpose were more color models developed.

To be suitable for image processing , a color space should meet the following criteria:

  • Compact storage of relevant color information (a clear separation of color and brightness information ).
  • Quick and easy conversion to and from other color spaces.
  • No jump points or singularities that make segmentation of the data difficult (perceptual models have a jump point in the color tone at 360 °).
  • Simple segmentation of regions based on color information.

The latter aspect was examined more closely by Ohta (1985). Ohta divided images into regions of similar color information using the histograms . Ohta experimentally determined the color features I1, I2 and I3 to be particularly effective for segmentation , which resulted in a new color space that is particularly suitable for such image processing processes. The following matrix shows the conversion from RGB to the I1I2I3 color space :

The I1 channel corresponds to the gray value axis of the RGB model. The channels I2 and I3, however, contain the color information.

Source

W. Hafner: Segmentation of video image sequences by adaptive color classification . Dissertation at the Institute for Computer Science at the Technical University of Munich, 1999, pp. 56–58.

Y. Ohta: Knowledge-based Interpretation of Outdoor Natural Color Scenes . Pitman Advanced Publishing Program, Boston / London / Melbourne 1985, pp. 12ff.

See also

  • Color space (further color spaces, color space systems, color models)

Web links

Individual evidence

  1. Hafner, 1999