Light field imaging technology provides users with a higher quality immersive visual experience.However,since it captures a vast amount of spatial and angular information,effective encoding methods are required to compress the massive data volume in light field images.In addition to focusing on compression efficiency,other important factors such as random access and viewport scalability need to be considered in light field image coding methods.Given these requirements,a novel light field image scalable coding method for light field images is proposed,which simultaneously takes into account the coding efficiency as well as the performance of random access and scalability.Firstly,the optical field image sub aperture array is sparsely sampled and the fifth viewport layer is discarded at the coding end.Then,based on the view hierarchical strategy,different multiple reference frame prediction relationships are conducted for encoding the viewpoints in the first four viewport layers.At the decoding end,viewpoint synthesis network is employed to synthesize viewpoints in the fifth layer,achieving compression and viewport scalability for light field images.The experimental results demonstrate that compared to the JPEG Pleno Anchor proposed in the JPEG Pleno general testing standard,the BDBR in the real scene is reduced by 16.166%,the BDBR in the synthetic scenes is reduced by 6.796%,and provides a more effective random access performance,with a relative random access penalty of the real scene at a bit rate of 0.75 bpp 18.91% on average and the relative random access penalty for synthetic scenes is 17.00% on average.The proposed method achieves a better balance between coding efficiency and random access performance.