Image based steganography uses the image as the cover media to overlay the payload. Spatial domain and frequency domain are the two familiar steganography techniques to carry embedding process on the cover image. Least Significant Bit (LSB) is commonly used spatial domain technique which is classified in to LSB replacement, LSB matching and matrix embedding. Further the embedding process may be sequential or random. This paper model an algorithm which give less distortion to the cover image during embedding process. Minimizing embedding impact and maximizing embedding capacity were few key factors of any steganography algorithm.  Higher the Peak Signal to Noise Ratio (PSNR) is lower the distortion. This is the fundamental principle followed in discriminating the distorted image (stego image) and cover image. Here matrix embedding technique is chosen to embed the secret image which is Huffman encoded. The Huffman encoded secret image is overlaid on the least significant bit of cover image through matrix embedding. As a result, the stego image is constructed with very less distortion when compared to the cover image. The stego image ends up with higher PSNR value. A secret image which cannot be embedded in a normal LSB embedding technique can be overlaid in this proposed technique since the secret image is Huffman encoded. PSNR value for different cover images and stego image with higher PSNR value is shown in this paper.