提出了基于快速响应分解与颜色空间转换的光学彩色图像无损加密算法。将彩色图像从RGB 空间转换HSI到颜色空间, 提取H、S、I三个成分; 根据随机相位掩码, 对H、S、I分量进行编码与调制处理; 基于Gyrator变换, 对相位编码后的H、S、I分量进行频域变换, 通过广义卷积操作, 形成一个Gyrator频谱; 引入相位-幅度截断机制, 得到Gyrator频谱的相位与幅度信息, 将幅度部分作为密文。为了实现无损解密, 利用快速响应分解技术对输出密文进行分割, 获取其相应的Q、R成分, 借助Gyrator变换对二者进行独立处理, 形成编码信息; 利用输出的相位信息来生成H、S、I分量对应的解密密钥, 通过联合编码信息与解密密钥, 获取HIS图像, 再将其转换到RGB空间, 实现无损解密。实验结果表明: 与当前彩色图像加密技术相比, 所提算法具有更高的安全性, 能够更好地抗击剪切攻击与明文攻击, 以及更高的解密质量, 很好地保留了初始明文的直方图特性, 视觉失真度最低。

The optical color image lossless encryption algorithm based on quick response decomposition and color space conversion was proposed in this paper. The color image was transformed from RGB space to HIS color space, and three components of H, S and I were extracted. The H, S and I components are coded and modulated according to the random phase mask. The phase coded H, S and I components were transformed in frequency domain based on the Gyrator transform, and Gyrator spectrum was formed by generalized convolution operation. The phase and amplitude information of Gyrator spectrum were obtained by introducing the phase -amplitude truncation mechanism, and amplitude information was regarded as the output cipher. In order to achieve lossless decryption, the corresponding Q and R components were obtained by using the quick response decomposition technique to segment the output cipher, and the encoding information was formed with the help of Gyrator transform to process the two codes. The decryption keys corresponding to H, S and I components were generated by using the output phase information and the HIS image was obtained by jointing encoding information and decryption key to realize lossless decryption for converting it into RGB space. The experimental results show that the proposed algorithm has higher security which can better resist shearing attack and plaintext attack, and the higher the decryption quality, as well as the lowest visual distortion which better preserving the histogram characteristics of the initial plain compared with the current optical color image encryption technology.