* Corresponding author. Tel.: #886-3463-8800; fax: #886- 3463-9357. E-mail address: csc#in@saturn.yzu.edu.tw (C.-F. Lin). Pattern Recognition 34 (2001) 671}683 Image hiding by optimal LSB substitution and genetic algorithm Ran-Zan Wang, Chi-Fang Lin*, Ja-Chen Lin Department of Computer and Information Science, National Chiao Tung University, Hsinchu, 300, Taiwan, Republic of China Institute of Electrical Engineering and Computer Engineering and Science, Yuan-Ze University, 135 Yuan-Tung Road, Chungli, Taoyuan, 320, Taiwan, Republic of China Received 29 September 1998; received in revised form 23 April 1999; accepted 24 November 1999 Abstract In this paper, we develop a novel method to embed important data in the host image so that the interceptors will not notice about the existence of the data. The basic concept of the proposed method is by simple LSB substitution. To prevent illicit access of data and increase the system performance, the method of randomized process and the optimal LSB substitution are proposed. The e!ectiveness of the optimal LSB substitution in the worst case is proved. We also develop a genetic algorithm to solve the problem of hiding important data in the rightmost k LSBs of the host image, which may involve a huge computation time to "nd the optimal result when k is large. Finally, based on the concept of perceptual modeling, an improved hiding technique is developed to obtain a high-quality embedding result. Experi- mental results will reveal that the quality of the embedding result will not be seriously degraded, and is near optimal by utilizing the proposed method.  2001 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved. Keywords: Image encryption; Image hiding; LSB substitution; Genetic algorithm 1. Introduction The popularity of the Internet o!ers a great conveni- ence to the transmission of a large amount of data via networks. However, it also increases the risk of illegal access and unauthorized tampering with content while transmitting data. To safely transmit data through the Internet networks, some mechanisms must be provided to guard important data against illegal interception. In the past decade, numerous techniques were proposed to provide the environment for important data transmis- sion. One of the most important techniques is by data encryption [1}4], in which the data are protected from illicit access by utilizing a particular cipher algorithm. The data are transformed by the cipher algorithm into a ciphertext that looks like a stream of meaningless codes, and then sent to the receiver via the network. Upon receiving the ciphertext, the receiver can decrypt it using a cryptography key obtained from the transmitter to recover its original form. Another approach to trans- mitting data safely, however, whose goal is quite di!erent from that of data encryption, is by data hiding [5]. Data hiding is the technique of embedding data into audio or video media called host signal such as music, images, and movies. It can be utilized in a wide variety of applications such as ownership identi"cation, tamper proo"ng, cap- tion, and secret data transmission. Unlike the goal of data encryption that prevents the data from illicit access and modi"cation, the aim of data hiding is to make the data inaudible or invisible to the grabbers. That is, the grabbers will not notice the existence of the embedded data, though they may listen very hard to or look very carefully at the media where the important data are hidden. 0031-3203/01/$20.00  2001 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved. PII: S 0 0 3 1 - 3 2 0 3 ( 0 0 ) 0 0 0 1 5 - 7