In Content Based Image Retrieval system images from the database are processed and their features are extracted. These features are such that they describe some properties of the image. The features of the images are indexed to determine the similarity between them. However, the retrieved image is the one having maximum similarity with the input image. Several research papers were studied in order to understand the evolution and evaluation techniques of CBIR. In this paper we present the pixel level features for content based image retrieval. We classify the feature extraction methods used and identify some promising techniques for Content Based Image Retrieval according to standard principles.