Steganography is the process of hiding secret information in other multimedia files, like image, audio and video, such that any unauthorized person does not have any clue of the hidden secret data. Due to the recent computational advancements, there has been an increased risk in secure communication and thus, steganography has become one of the most researched domains in information security. In this paper, a novel method of steganography is approached where the cover image is scrambled using graph theory; then the secret data is embedded in the LH, HL and HH planes using LSB embedding of the cover image after applying Haar forward transform to get the embedded image. The secret data hidden in the cover image is first converted to a QR-like (Quick Response) image and then to be embedded in the cover media. The stego image is obtained after reverse Haar transform and descrambling of the embedded image. The proposed method when applied to a set of grayscale images, not only gave us stego images with greater embedding capacity but also high Peak Signal to Noise Ratio (PSNR) and a low Mean Square Error (MSE) and very appreciable Normalized Cross-Correlation (NCC) along with a good Structural Similarity Index Matrix (SSIM) when compared to the cover images.