Content-based image retrieval (CBIR) is an image retrieval technique used to retrieve images efficiently by using low level features (like texture, shape, color) of those images. It is an efficient retrieval mechanism to retrieve images from the multimedia database. The combination of these low level features give better results than individual feature. In the previous years, the amount of low level features to be retrieved were decided based on manual assumption or heuristic experiments. In this paper, we have focused on features extraction process dynamically and improving the precision value. The weightage of image features (color and Shape) are calculated dynamically and then their combination is used to retrieve more relevant images. Euclidean distance method is used for feature matching. An efficiency of image retrieval is measured by precision value of it.