<p>With the development of the information age, the security of image data has become increasingly important. To enhance image security in applications such as data hiding and privacy protection, this study proposes an innovative encryption framework that synergizes a random Zigzag scrambling scheme with a roulette wheel rotation diffusion method. The system constructs a novel four-dimensional hyperchaotic model by embedding an additional state variable into a classical three-dimensional chaotic structure. The simulation results confirm that the generated chaotic sequences exhibit strong hyperchaotic behavior and uniformity across diverse parameter settings. The encryption approach effectively disrupts both the pixel positions and intensity values through combined Zigzag path scrambling and rotational diffusion, both of which are deeply coupled with the designed chaotic system. Notably, the chaotic system's initial parameters are computed directly from the grayscale values of the source image, ensuring that encryption is highly sensitive to both the plaintext and the key, thus strengthening security. The performance evaluation indicates that the method offers an expansive key space on the scale of 2³⁸⁴ × 10⁶⁰, while the entropy of the encrypted image reaches 7.9993 bits per pixel. Furthermore, the algorithm achieves a normalized pixel change rate (NPCR) of 99.6048% and a unified average changing intensity (UACI) of 33.4706%, demonstrating strong resistance to statistical attacks and robustness against cropping and noise-based tampering. Overall, the method delivers a high level of encryption performance with notable resilience.</p>