In this paper, a novel method for automatically detecting High Dynamic Range (HDR) scenes is proposed along with a method for determining suitable frames for fusion. The proposed algorithm finds its application in multi-exposure image fusion systems, such as consumer digital cameras etc. And, it is of interest to minimize computation by avoiding redundant image fusion. The proposed algorithm makes use of two or more auto-exposure bracketed frames to determine the degree of details in form of Momentum Value (MV). MV is computed region-wise for each frame when available. After accumulating MVs for all bracketed frames, they are compared regionwise to select frames that captures maximum information. If more than one frame is required to get maximum MV for all regions, then the scene can be judged to have dynamic range higher than that of the camera. And, the frames with maximum MV are used for fusion. The proposed approach consumes approximately 2.5% of the computation efforts compared to+ image fusion. Effectively, proposed method leads to significant saving in computations and enhances quality of fused HDR image.