Color images play an essential role in computer vision systems where colors are useful information. Extraction of color images features is the backbone of many pattern recognition and image processing applications such classification of color images, color image retrieval, color image watermarking and template matching. Moments and moment invariants are widely used as global and local features for gray images. Quaternion moments are used in representation of color images. In this chapter, an overview of quaternion moments and their rotation, scaling and translation invariants is given. A series of numerical experiments are performed using different parameters to assess the performance of extracted features using different quaternion moments.