Method and apparatus for representing a group of images

A method of representing a group of images using a processor to process signals corresponding to the images, the method comprising:

• obtaining, using a processor, a dominant colour representation for each image in the group of images, wherein each of the dominant colour representations has one or more components including at least one dominant colour value, and optionally for each dominant colour value, a variance indicating colour distribution variance of the image with respect to the dominant colour value, a weight indicating the influence of the dominant colour in the image, the number of dominant colours in the image, and spatial homogeneity of pixels corresponding to the dominant colors in the image;
• determining the values of one or more dominant colours for the group of images;
• combining the components including at least the dominant color values of the dominant colour representations to determine combined components including at least a combined dominant colour value for the group of images;
• deriving, using a processor, a group dominant colour representation expressing the group of images in terms of said combined components including at least said combined dominant colour value for the group of images;
• and
• storing the group dominant colour representation as a descriptor for the group of images,
• wherein for two dominant colours m1 and m2, and respective weights W1 and W2, the merged dominant colour m has the value:
• m=w1m1+w2m2,
• where w1, w2 are the relative weights, W1/W1+W2 and W2/W1+W2 respectively.