可回復式高動態範圍影像資訊隱藏演算法

Abstract

本論文提出一個應用在高動態範圍影像上的可回復式資訊隱藏演算法，就我們所知，此演算法是文獻上的首創。我們的方法具備可回復式的特性，即在取出秘密訊息後還可完整的重建原來的高動態範圍掩護影像。演算法亦藉由可適應性方式嵌入秘密訊息來達到高嵌入量的目的。本篇論文利用各色彩頻道的差值結合相鄰E值差異進行可適應性的判斷，因此可以達到更高的嵌入量，同時具有還原掩護影像的特性。實驗結果顯示：我們的演算法在單次嵌入的情況下，平均每像素之嵌入量可達0.13~0.63位元，其NRMSE數值為0.05~0.83%，經過色調再生處理後，其平均PSNR數值仍能維持在39dB以上，具備極佳的視覺品質。This paper proposes a reversible data hiding algorithm for high dynamic range (HDR) images. To the best of our knowledge, our scheme is the first reversible approach, which can recover the cover HDR image without any distortion after the secret data being extracted. Our algorithm takes advantages of the differences of the neighboring pixels in the R, G and B channel to provide large embedding capacity. We exploit the difference of the neighboring pixels in the exponent channel E to embed dynamic ranges of the secret message, achieving that embedding adaptability. Experimental results show the our algorithm provides an average payload capacity for up to 0.13~0.63 bits per pixel (bpp). In addition, our algorithm produces a stego image with high visual quality. The normalized root mean squared error (NRMSE) is 0.05~0.83%, and the peak signal-to-noise ratio (PSNR) between the stego and cover HDR images after the tone mapping process is over 39 dB. Experimental results verify the feasibility of our algorithm in providing a reversible solution for steganography on high dynamic range images.