Steganography is the art and science of hiding secret data within a carrier medium, such as digital images, without altering its perceptible quality. In this research, we introduce an advanced steganography technique that combines SLIC segmentation with Quaternion Wavelet Transform (QWT) to securely and robustly embed encrypted messages into images. Our steganography technique overcomes several drawbacks of traditional methods such as noticeable visual distortions, and susceptibility to various attacks. Initially, the image is segmented into foreground and background regions using SLIC, ensuring that the embedding process respects the perceptual importance of different areas within the image. We then apply QWT to decompose the image into approximation and detail coefficients. The secret message undergoes dual encryption using both substitution and transposition ciphers, significantly improving its confidentiality. This encrypted message is subsequently converted into binary form and embedded into the approximation coefficients of the wavelet-transformed image. To evaluate the efficiency of our method, we computed PSNR (Peak Signal-to-Noise Ratio), MSE (Mean Square Error), and SSIM (Structural Similarity Index) metrics, demonstrating superior performance compared to traditional techniques such as DCT (Discrete Cosine Transform), DWT (Discrete Wavelet Transform), and LSB (Least Significant Bit) methods. Additionally, we tested the robustness of our steganographic images against various attacks, including Gaussian noise, salt-and-pepper noise, rotation, translation, blurring, and compression, using Bit Correct Rate (BCR) as the evaluation metric. Our results show that the proposed method achieves BCR values exceeding 94%.