This manuscript introduces a novel self-adaptive image encryption scheme that leverages chaos theory and gravitational models to aggrandize the security of image transmission. The approach features a key generation mechanism based on bit planes and is closely tied to the plaintext, ensuring its self-adaptive nature. It operates in three main stages: key-generation, double scrambling, and double diffusion. The processes involve mixed chaotic transforms, shift row transforms, diagonal scanning transforms, and XOR operations to achieve effective scrambling and diffusion. Simulation and investigative findings indicate that the proposed scheme offers robust encryption and decryption capabilities, demonstrating strong resistance to various attacks. It outperforms several recently proposed algorithms in terms of security. This encryption framework is particularly suited for secure image transmission and storage applications. The scheme extends traditional encryption structures to bolster security, with scrambling and diffusion being efficiently executed through diagonal scanning transforms and XOR operations. By incorporating Two-Dimensional Logistic Tent Map(2D-LTMM) and One-Dimensional Sine Chaotic System(1D-SCS) chaotic systems, which exhibit ultra-high chaos performance and low computational cost, the scheme significantly enhances randomness. Overall, the results confirm the scheme’s superior performance in both encryption and decryption, alongside its enhanced security features.