Concerning the problem that the traditional image compression ratio control is rough and the low-dimensional chaotic system has low confidentiality, the image compression encryption algorithm based on compression ratio control of Joint Photographic Experts Group-Lossless Standard (JPEG-LS) was proposed. On the basis of further analysis about the distortion control parameter Near of JPEG-LS which had significant influence on the image compression ratio and reconstruction quality, the image data of raster scan was first gradient processed in causal model, the size of the gradient value and the Near were compared to determine the run-length coding in run mode or Golomb coding in normal mode, and then random processing sequence generated by three-dimensional Lorenz chaotic system was regarded as the key to encrypt the compressed code stream in run mode, conventional model and whole model (run and conventional model). Finally, the fine control of image compression ratio and improvement of the confidentiality were realized by dynamically adjusting Near in real-time. The simulation results indicate that the proposed algorithm has a good performance on control of the compression ratio, and its quality of reconstruction image is improved about by 0.5 dB than that of linear compression ratio control algorithm. Furthermore, it has high security that can efficiently resist entropy attack, differential attack, brute force attack and statistical attack, etc., and the encryption has no effect on the compression efficiency.