Inverted Pattern Approach to Improve Image Quality of

Introduction (1/2) n Simple LSB substitution 1 0 1 0 1 (181 ) Data

Introduction (2/2) n Optimal pixel adjustment process (OPAP) K bits A pixel H :

Proposed scheme (1/2) n Inverted pattern approach 1 0 0 1 (inverted of S)

Proposed scheme (2/2) n Inverted pattern approach We could use 1 bits per block

The experiment results (2/4) cover image The results of embedding 512 × 256 secret

The experiment results (3/4) cover image The results of embedding 512 × 256 secret

The experiment results (4/4) The MSE values of embedding 512 × 256 secret images

Conclusions Short computation time n High capacity n Better quality than previous methods n

Slides: 10

Download presentation

Inverted Pattern Approach to Improve Image Quality of Information Hiding by LSB Substitution Source: Pattern Recognition, In Press, Accepted Manuscript, Available online 9 February 2008 Authors: Cheng-Hsing Yang Speaker: Pei-Yan Pai Date: 2008. 3. 11 1

Introduction (1/2) n Simple LSB substitution 1 0 1 0 1 (181 ) Data string: 0010101100011101010011… 1 0 1 1 0 0 1 0 (178 ) 2

Introduction (2/2) n Optimal pixel adjustment process (OPAP) K bits A pixel H : MSB A K-bit string S, If LSB – S > Else If LSB – S > LSB i. e. , H =10110101 (181) , H’=10111110 (190), k=4 => LSB=0101 (5), S =1110 (14) , then LSB – S =5 -14= -9 is greater than So, 3

Proposed scheme (1/2) n Inverted pattern approach 1 0 0 1 (inverted of S) Split the embeded bit string to P parts, P=pattern length i. e. , 11010000111 P=2 P=4 P=8 11010110 1101 0110 10000111 1000 0111 11 01 01 10 10 00 01 11 4

Proposed scheme (2/2) n Inverted pattern approach We could use 1 bits per block to represent inverted or not inverted. i. e. (1= inverted, 0 = not inverted) Original string: 11 01 01 10 10 00 01 11 Modified string: 11 10 10 10 01 00 01 11 Inverted 0 pattern: The embedding rules: 1 1 0 0 0 The embedded string is S, and the replaced string is R For P part in i =1 to P i. e. s=111 =000 P: 10010101(149) If MSE(Si, Ri) MSE( i, Ri) Ps: 10010111 (151) P : 10010000 (144) Choosed Si for embeded MSEPs = 4 MSEP = 25 If MSE(Si, Ri) Choosed End i MSE( i, Ri) for embeded So, choose s for embedding. 5

The experiment results (1/4) 6

The experiment results (2/4) cover image The results of embedding 512 × 256 secret images into cover image Lena by different LSB substitution approaches: OPAP LSB and IP LSB with different pattern lengths. 7

The experiment results (3/4) cover image The results of embedding 512 × 256 secret images into cover image Baboon by different LSB substitution approaches: OPAP LSB and IP LSB with different pattern lengths. 8

The experiment results (4/4) The MSE values of embedding 512 × 256 secret images into 512 × 512 cover images by different approach. 9

Conclusions Short computation time n High capacity n Better quality than previous methods n 10