This paper proposes a hybrid picture encryption technique that combines the effectiveness of a deep learning-based autoencoder with the unstructured properties of the logistic map. The logistic map is utilized for encryption, ensuring robust security through its chaotic behavior and sensitivity to initial conditions. Additionally, an autoencoder architecture is employed for image transformation, facilitating efficient encryption and decryption processes while preserving image quality. An experimental evaluation demonstrates the effectiveness of the suggested method in terms of computing efficiency and security. The encryption scheme exhibits resilience against various cryptographic attacks, while the autoencoder enables fast processing and maintains image fidelity. Overall, the hybrid encryption approach offers a promising solution for secure image communication and storage in diverse applications, including medical imaging and confidential data transmission. The fusion of deep learning-driven transformation and chaos-based encryption achieves a harmonious equilibrium between security and performance, rendering it well-suited for practical applications in encryption.