<p>This paper proposes a joint multi-image encryption framework for images of diverse types and sizes, which integrates a novel two-dimensional chaotic system Sine-Logistic-Exponent map(2D-SLEM) with three-dimensional spatial processing to achieve multi-image encryption. Based on the chaotic characteristics of Sine map and Logistic map, the exponential function is integrated and extended to two dimensions, and a new chaotic map 2D-SLEM is proposed. The mean sample entropy can reach 2.192. In the preprocessing stage, the high-bit and low-bit XOR diffusion processing is performed on the pixel values of images of different sizes. The diffused values are stored and concatenated to form a pixel value library, enabling the extension and unification of different image sizes. The preprocessed images are combined with SHA-512 to generate initial key parameters. The chaotic sequence is generated by iterative 2D-SLEM. Based on the chaotic sequence, the index is generated and multiple images are mapped into a three-dimensional cube image. Inspired by two-dimensional planar block processing, the cubic image is partitioned into three-dimensional blocks. A novel pixel three-dimensional spatial jumping rule is proposed. Subsequently, a dynamic sliding circular queue is employed to store binary pixel values along different axes of the three-dimensional image under the control of chaotic index sequences. By managing the sliding of dynamic circular queues and the entry and exit of elements, and XOR operation with chaotic sequence, nonlinear diffusion of multi-channel pixel values in three-dimensional space is realized. Finally, through the simulation and performance analysis of the encryption scheme, the average value of the image information entropy after multi-size image encryption can reach 7.9994. NPCR and UACI can reach 99.6153% and 33.4576%. The histogram shows a uniform distribution feature, indicating that the encryption method has strong robustness and security.</p>