植基於邊緣預測之可逆式隱藏技術

Reversible data hiding based on side-match prediction

蔡蕙如(Huei-ju Tsai)

指導教授 ： 李金鳳

朝陽科技大學/資訊學院/資訊管理系/碩士(2011年)

摘要

資訊隱藏就是將機密訊息嵌入在一個媒體中，媒體可以是聲音、影像或影音等等。資訊隱藏技術的優點就是讓經加密的媒體具備偽裝作用，讓人類感知系統無法察覺媒體有異並進而達到資訊安全的效用。早期不可逆式資訊隱藏技術有高藏量之優點，可將大量機密訊息嵌入至媒體中，收方可取回機密訊息但遭破壞的媒體卻已無法還原，因此這類不可逆式資訊隱藏技術將不適用於醫療、軍事或天文等特殊領域。為供應特殊領域所需，近年來許多可逆式資訊隱藏技術紛紛崛起，讓嵌入媒體中的機密訊息不但能取出且遭破壞的媒體也能予以還原。除此之外，可逆式資訊隱藏技術會為了保有影像可還原之依據而衍生出額外資訊，這些額外資訊量會佔據些許總藏量的空間，故影像中真正藏入實際機密訊息的量往往會不如不可逆式資訊隱藏技術。從過去到現在，不少學者皆採用邊緣之像素值進行預測，主要是因為利用邊緣像素值來進行預測能有助於尋找影像特徵，因此學者們可依不同特徵設計出最適化之嵌入技術。然而利用邊緣預測的技術中，擁有高藏量之特色的技術多為不可逆式資訊隱藏技術。本研究提出一項邊緣吻合預測之可逆式資訊隱藏法，可應用在醫療或軍事用之影像上，除了應用較廣泛之外，藏量平均可達2.05bpp，是一個藏量相當高的可逆式資訊隱藏法。雖然本研究追求高藏量的同時，會產生些許的額外資訊，但是本研究將這些額外資訊進行無失真壓縮後再嵌入至影像中，因此省去需額外傳輸的負擔，且平均每張影像品質仍能維持在42.22dB左右。為了要處理額外資訊，勢必犧牲些許的影像品質，若能減少額外資訊的產生便能換得更高之影像品質，因此本研究提出可逆式多鄰近像素值預測隱藏法，利用多個像素值來搭配線性預測，除了有助於影像獲得較佳的預測外，嵌入期間產生的額外資訊因為十分少量，所以不需犧牲影像品質來處理這些額外資訊，故影像品質平均都能維持在48.65dB左右，具絕佳的視覺品質。

關鍵字

可逆式資訊隱藏技術 ； 直方圖位移操作 ； 邊緣預測 ； 預測誤差

並列摘要

Data hiding is the technique to embed information into a media, which could be sound, picture, or video file, etc. The advantage of data hiding technique provides camouflage ability for encrypted media, thus human perception system could not recognize the modification made to the media and achieve information security. Early versions of irreversible data hiding technique stands out with high capacity, so large amounts of secret messages could be embedded into media. Yet it would be impossible for the receiver to recover the damaged media after retrieving the secret messages, hence this irreversible technique is not suitable for specialized fields such as medical, military, and astrological. To fulfill the needs of specialized fields, many reversible data hiding techniques showed up recently, so the damaged media can be recovered after the embedded secret messages have been retrieved.
In order to retain the image recovery for reversible data hiding technique, some extra information will be produced. This overhead data will consume the embeddable spaces. Therefore, the pure payload of reversible data hiding is always less than that of irreversible data hiding technique. From the past until the present, many researchers used side pixel values to perform prediction since these values facilitate the search for image characteristics. Therefore, researchers could design the better embedding technique according to various image characteristics. The techniques that possess high capacity are mostly irreversible data hiding techniques.
This study proposes a reversible method of data hiding using side match prediction. Even though it also utilizes side prediction to embed, this proposed method can be applied among medical and military images. Despite having wide applications, the proposed method has a payload of 2.05bpp on average, which is considered as a reversible data hiding technique with extremely high capacity. Some extra information will be generated as this study pursues high capacity. However, the extra information was embedded in images after lossless compression in this study, thus burden from extra transmission is avoided while an average of 42.22dB of image quality is maintained.
In this study, certain image quality has to be sacrificed to process extra information. If the generation of extra information could be reduced, a higher image quality can be obtained. Thus, the method of prediction-based reversible data hiding using similarity of neighboring pixels is proposed in this study by using multiple pixel values to accompany linear predictor. Besides providing a better prediction of images, quality of images will not be sacrificed to process extra information since the amount of it is extremely small in such mechanism. As a result, image quality is maintained at around 48.65dB on average with excellent visual quality.

並列關鍵字

Histogram shifting ； Reversible data hiding ； Side match prediction ； Prediction error

參考文獻

[2] C. C. Chang and H. W. Tseng (2004), "A steganographic method for digital images using side match," Pattern Recognition Letters, vol. 25, pp. 1431-1437.

[3] C. C. Chang, C. Y. Lin, and Y. Z. Wang (2006), "New image steganographic methods using Run-Length approach," Information Sciences, vol. 176, pp. 3393-3408.

[4] C. C. Lin, W. L. Tai, and C.C. Chang (2008), "Multilevel reversible data hiding based on histogram modification of difference images," Pattern Recognition, vol. 41, pp. 3582-3591.

[5] C. K. Chan and L. M. Cheng (2004), "Hiding data in images by simple LSB substitution," Pattern Recognition, vol. 37, pp. 469-474.

[6] C. N. Lin, D. J. Buehrer, C. C. Chang, and T. C. Lu (2010), "Using quad smoothness to efficiently control capacity-distortion of reversible data hiding," Journal of Systems and Software, vol. 83, pp. 1805-1812.