Steganography plays an important role in information sharing. In this paper a wavelet based image steganography using Haar wavelet and an N-ary notation is proposed. The high and middle frequency coefficients are selected based on the Shannon entropy value to identify the best sub band for data hiding. Peak signal to noise ratio and Structure similarity index metric values are computed from the stego image. The empirical results show the proposed method provides good, better imperceptibility and data capacity with average peak signal to noise ratio 53.60 dB for 359,112 bits in terms of effectiveness. To prove this accomplishment of the method, several experiments were conducted and compared the results with existing works. Copyright © 2019 Inderscience Enterprises Ltd.