As digital communication and multimedia data exchange develops at an extremely fast pace, the importance of confidentiality and integrity of the information relayed has become even more important. Images are one of the widely used images in various fields like healthcare, defense, and cloud computing, which are highly susceptible to unauthorized access and data breaches. In this paper, a safe data hiding model founded on the Discrete Wavelet Transform (DWT) methods is proposed to overcome these drawbacks. The main goal is to come up with an adaptive and efficient embedding system that can be highly imperceptible and yet high security. Wavelet decomposition is proposed to split the image into sub-bands (LL, LH, HL, HH) and secret data is concealed in a few high-frequency bands with the help of a pseudo-random approach based on the key. A selection method based on adaptive coefficient is used to select the best positions to embed and reduce distortion. Standard benchmark images are used to test experimental performance and the outcomes are evaluated in terms of Peak Signal to Noise Ratio (PSNR), Mean Square error (MSE), Structural similarity Index (SSIM) and Payload capacity. The findings show that the suggested technique has a better performance with PSNR values of over 44 dB, MSE of about 1.3, SSIM of more than 0.98, and a payload capacity of up to 0.4 bpp. A comparative study reveals that the method has great advancements over the traditional and current DWT-based methods in terms of image quality and robustness.