In this paper, a depth-based steganographic framework for robust and transparent message hiding in 3D stereo images is presented for a 3D image sequence. The framework uses high-accuracy depth maps from Middlebury stereo datasets (2014 and 2021) and combines zero-shot object segmentation with the Segment Anything Model (SAM), and encodes secret messages using Least Significant Bit (LSB) substitution. For imperceptibility validation, We employ Depth-Image-Based Rendering (DIBR) to render new views from stego depth maps ans uses PSNR, SSIM as well as VIFP metrics to gauge the quality. Despite the random corruption policy, our method achieves more than 35 dB PSNR, 0.95 SSIM and 0.85 VIFP while making fewer recovery errors compared with the other methods for next frame video SR. The proposed framework overcomes the limitations of conventional pixel-based steganography by offering low detectability and suitability for 3D multimedia systems like virtual and augmented reality.