Reversible Color Image Steganographic Scheme based on Shape-specific point

(植基於圖型特徵點之可逆式彩色影像隱寫術)

 

Ching-Yu Yang

Dept. of Comp. Sci. and Inform. Eng., National Penghu University of Science and Technology, Taiwan

No 300, Liu-Ho Rd., Magong, 880 Penghu, Taiwan

chingyu@npu.edu.tw

 

Wen-Fong Wang

Dept. of Comp. Sci. and Inform. Eng., National Yunlin University of Science and Technology, Taiwan

No 123, University Rd., Section 3, Douliou, 64002 Yunlin, Taiwan

wwf@yuntech.edu.tw

 

楊慶裕

國立澎湖科技大學資工系

880澎湖縣馬公市六合路300

chingyu@npu.edu.tw

 

王文楓

國立雲林科技大學資工系

64002雲林縣斗六市大學路3123

wwf@yuntech.edu.tw

 

 

摘要

植基於兩個圖型特徵點:半徑加權均值RWM)和中心值Centroid),本文使用RWM模設計出一個可逆式彩色影像隱寫技巧。換言之,我們使用區塊RWM模辨別策略,可有效的將密位元嵌入主體影像,並於接收端從標示影像無失真取出;而且可完全回復主體影像內容。值得一提的是本法也具備某種程度之強韌性質 (此特性是大多數傳統影像隱寫術所無法達到的)。實驗結果顯示本法產生的視覺品質不錯且具中等隱藏容量。使用本法產生的標示影像可抵抗彩色量化、(邊緣)銳化、高斯/均勻雜訊添加、像素反向、JPEGJPEG 2000和風阻等攻擊。也就是說,我們從被攻擊過的標示影像中所取出的版權圖像(或標誌仍可識別。

 

關鍵字: 資料隱藏、可逆式彩色影像隱寫術RWM、秘密通訊

 

Abstract

A reversible steganographic method for color images using two shape-specific points, namely, radius-weighted mean (RWM) and centroid is proposed. To embed data bits in a host block, the decision policy of the Euclidean norm of RWM (RWM-norm) is employed in the proposed method. The proposed method can restore host images after bit extraction with a certain degree of robustness. Simulations confirmed that the resultant perceived quality was good, while hiding capacity was moderate. Marked images generated by the proposed method could resist various attacks such as color quantization, (edge) sharpening, Gaussian/uniform noise addition, inversion, JPEG and JPEG 2000 compression, and winding attacks.

Keyword: Data Hiding, Reversible Color Image Steganography, RWM-norm, Covert Communications