摘要 |
A method of determining a unique number of colors for an image is described which uses homotopic transformation (in particular the watershed transformation) within feature space formed by red, green and blue components, or any other set of attributes, of the pixels of the image. Each color is given a label and the image is then segmented into regions of uniform labels. The color or characteristic of each segment can then be altered or used to identify an object. The method can be applied to any scalable, either integer or real, data set. |