Abstract: Considering the complexity of channels which constitute the high dynamic range（HDR） image, it is hard to hide information into HDR images. In this paper, a reversible data hiding algorithm for HDR images based on prediction-error expansion（PEE） is proposed. First, HDR images are encoded using radiance format. Second, the redundancy existing in channels is exploited to classify the pixels into flat and edge areas. Then, adopt two different embedding manners for different areas to embed data into pixels according to the sorted order. Finally, the cover HDR images can be fully recovered after data extraction. Various experiment results show that compared to traditional methods, the proposed method make better use of the redundancy among channels, and thus achieve high embedding capacity with low distortion in the stego image.