User:AndrewBurnie

Colour Edge Detection is a process in computer vision  for finding boundaries between areas in full colour images. There are many methods for Edge detection in grayscale images but in this colour space two very different colours may have the same intensity values. For applications of computer. There are three predominant types of colour edge detection system, output fusion, multi-dimensional gradient models and vector models. These could operate simply on the hue of areas of the image but this will miss any edges where the change is brightness so we need to use a combination of the two.

Output Fusion
thumb|400px|right|Steps of an output fusion algorithm Output fusion is one style of colour edge detection and there are many methods which use this structure. The colour image is first segmented into its red, green and blue components each of which are used to create separate gradient maps. Edges are then detected in these gradient maps using a standard edge detection method, such as the use of the Sobel operator and the three edge maps are then recombined to form the complete edge map.

This recombination is as simple as addition of the three edge maps or take the maximum of the three values at any pixel. Simple addition will give strong edges where every colour has an edge whereas taking a maximum will leave an edge map that has strong edges anywhere any one of the three edge maps does.

Multi-Dimensional Gradient Models
thumb|400px|right|Steps of a multi-dimensional gradient model algorithm Multi-dimensional gradient models short cut the output fusion process. While images are split into the three channels, red green and blue. Their individual gradient models are then recombined to create an overall gradient model which is used for edge detection. Again this combination can be done in a variety of ways yielding different results. Multi-dimensional gradient models are a system more similar to how humans perceive a colour scene but this may not be helpful in

Vector Models
Vector models attempt to find edges by treating pixel's colour as a vector and using these vectors to detect edges. These models have not become popular as yet

Colour Maps
It is possible to process images in another colour space such as HSI but this adds to the complexity as conversion has to be applied and this makes them unattractive