Wireless Capsule Endoscopy (WCE) generates a large number of images in one examination of a patient. It is very laborious and time-consuming to detect the WCE video, and so limits the wider application of WCE. Color similarity measurement is the key technique of color image segmentation and recognition, as well as the premise of bleeding detection in WCE images. This paper deduces two color vector similarity coefficients to measure the color similarity degree in RGB color space, and based on which, a novel method of intelligent bleeding detection in WCE image is implemented. The novel algorithm is implemented in RGB color space, and is featured with simple computation and practicability. The experiments showed that the bleeding regions in WCE images can be correctly extracted, and the sensitivity and specificity of this algorithm were 90% and 97% respectively.