Retrieval of images from large collections of image databases based on the image content has become an important issue for database and image processing communities. It is generally accepted that texture and color are the key features for the Image Retrieval Systems. Normally texture is extracted from gray scale images and the contribution of color to the perception of texture has been ignored mostly. The paper proposes a method which combines the human perception of color and texture employing supervised learning to obtain an overall impression of image. The initial results of the proposed technique found to be promising. Some of the results have been reported in the paper.