Histogram-based object recognition

Color Constancy M. Ebner 2007 John Wiley Sons, Ltd 

Each bucket //(c) is simply the probability that the color vector c occurs in the original image. 

Swain and Ballard have proposed a method called histogram intersection to compare two color distributions. Let /// be the histogram of the input image. Let Hm be the histogram of the model object. The given histogram is compared to the histogram of the model 

The intersection simply counts the number of pixels of a particular color in the model image, which are also found in the test image. In order to obtain a match value with the range [0, 1], the intersection is normalized by the total number of pixels in the model. 

If we are working with normalized histograms where the buckets contain the probability that a particular color is found in the image, then the normalization can be omitted because we have IIm (c) = 1. This particular method of comparing histograms was defined 

---