Abstract—Rapid development of information and communication technology has increased the need of protecting sensitive data from illegal access during their transmission. One of popular methods to secure such data is by embedding them in another file such as a digital image, which is called as data hiding. This involves processing values of pixels that causes image distortion. This noise has become one of common problems in data hiding, along with the capacity of the secret data, which can be carried by the image for example. Some methods have been introduced in order to alleviate these two problems. Nevertheless, this is still challenging. In this paper, we propose a new data hiding mechanism, which is developed based on the difference between neighboring pixels to construct pixel value grouping, called adaptive pixel value grouping (APVG). This scheme selects pixel group head which is then compares it with other pixel to get difference value. This difference determines how the pixels in the image are grouped. This resulted group, if any, is used for further embedding process. The experimental result shows that APVG scheme is able to achieve a better quality, which is depicted in PSNR value, than the existing method. This proposed method achieves 65 - 72dB, 54 – 61dB and 44 – 55dB of PSNR for 1kb, 10kb and 100kb payload (secret data), respectively.