Abstract: Digital color images are the most widely used data types and are used in many vital applications such as computer recognition systems and image retrieval system, so, creating an image features victor is an important issue, this victor can be used as a signature or key to identify or retrieve the image from a color image database. Two methods of color image features extraction will be proposed, implemented and analyzed, local contrast method and a proposed method which will be based on local binary pattern. The obtained experimental results will be compared in order to raise some recommendations which will be useful when building power image recognition systems.