Fund Project:Project supported by the National Natural Science Foundation of China (Grant Nos. 61565010, 11762009, 61865007) and the Natural Science Foundation of Yunnan Province, China (Grant No. 2018FB101).
Received Date:28 January 2019
Accepted Date:05 March 2019
Available Online:01 June 2019
Published Online:05 June 2019
Abstract:In this paper, the function of parallel encrypting multiple images and reproducing arbitrary layers of images is realized by improving the double pupil function in optical scanning holography. In an optical scanning holography (OSH) system, a dual-pupil heterodyne incoherent image processing technique is used to record holographic images. By adjusting the two pupil functions in the optical system, the interference fringes can be modified to achieve different imaging effects. In this paper, the ring pupil and random phase plate are used to act as two pupil functions to interfere to form a ring random phase plate, and thus realizing the fast scanning of multi-layer images. Then the multi-layer images can be quickly encrypted by one imaging technique. The scanned signals are quickly collected by photoelectric detectors, and they synthesize encrypted holograms by computer. By using the digital holography to decrypt the holograms, the precise reproduction of any layer image can be achieved. The OSH system with random phase pupil is strongly dependent on the longitudinal position of the system in digital reconstruction. The defocusing noise can be converted into random noise and the effect of defocusing layer on imaging can be effectively suppressed. However, in practice, it is necessary to average multiple images to achieve better imaging effect, and the accuracy of random phase plate is required. In this paper, most of the random noise can be filtered with the aid of ring pupil, and all the information about multi-layer graphics can be recorded and reconstructed by one scan. In the process of reconstruction, the influence of defocusing image can be effectively eliminated, and the decryption of any layer image can be realized. This method collects encrypted image by photoelectric detector, and does not need complex algorithm reconstruction nor phase iteration, which greatly reduces the time expended in the encryption process. In the process of encryption, the key space of the system is increased, and the encrypted image obtained has high security. In this paper, correlation coefficient is used to evaluate the encryption effect of this method, and the effectiveness and security of this method are verified by simulation experiments. For cutting resistance, when 75% of the information is lost, the correlation coefficient can still reach more than 0.5. For the sensitivity of information, the integrity of decrypted image will be seriously damaged when the wavelength and distance shift very little. For the anti-noise ability, under the influence of Gauss noise and salt and pepper noise, the correlation coefficient and image recognition degree are high. This method is very time-saving, and the result of encryption has high security, high sensitivity, strong ability to resist clipping and noise. Keywords:optical scanning holography/ optical parallel encryption/ annular pupil/ random phase plate
光学加密系统的敏感性分析是评估系统抗攻击能力的重要环节. 本节的实验中, 只对切片1 (z1 = 10 mm)进行数字重建的对比实验. 在光学加密系统的解密过程中, 重建距离、波长等受到干扰时, 会在一定范围内浮动. 对于重建距离, 假设重建时的距离在一个很小的范围内浮动, 结果如图7(a)—图7(c)所示. 图7(a)是相关系数值随距离变化的曲线, δd是重建距离浮动的大小, 即δd = |zc–z1|, 其中图7(a)和图7(d)中的相关系数是与图6(a)的正确解密结果相对比得到的. 图7(b)和图7(c)分别是δd = 0.01 mm和δd = 0.1 mm时的重建解密图像, 相应的相关系数分别为0.1985和0.0596. 可以发现, 当距离浮动0.01 mm时相关系数很低, 图像的信息已经几乎无法识别, 而错误距离为0.1 mm时, 图像信息已经完全无法识别. 对于重建波长, 如图7(d)—图7(f)所示, 图7(d) 是相关系数值随波长变化的曲线, δλ是重建波长浮动的大小. 图7(b)和图7(c)分别是δλ = 0.1 nm和δλ = 1 nm时的重建解密图像, 和重建距离错误类似, 但是波长的敏感性比成像距离更高, 重建波长浮动的距离为1 nm时, 相应的相关系数为0.0049, 此时图像信息就已经完全无法识别, 系统的敏感性很高. 所以当解密过程中的任何一个参数错误时, 都无法从加密图像中解密出原图像的任何信息, 而且即使攻击者破解了其中一层的密钥信息, 但是由于相邻两层加密图像的位置信息都不同, 且没有规律性(每两个切片之间的距离不同), 所以也无法通过任何一个切片的密钥解密出其他切片的图像信息, 可以有效地抵抗暴力攻击. 图 7 错误解密结果 (a) 相关系数Cc随δd变化的曲线图; (b) 和 (c) δd = 0.01和0.1 mm时的解密结果; (d) 相关系数Cc随δλ变化的曲线图; (e) 和 (f) δλ = 0.1和1 nm时的解密结果 Figure7. Incorrect decryption results: (a) The Cc curve of varies with δd; (b) and (c) decryption result of δd = 0.1 and 0.1 mm, respectively; (d) the Cc curve of varies with δλ; (e) and (f) decryption result of δλ = 0.1 and 1 nm, respectively.
23.4.系统抗剪裁性能分析 -->
3.4.系统抗剪裁性能分析
在解密原理中, 加密图像的像素和解密图像的像素之间存在着一一对应的关系, 也就是说, 加密图像的部分丢失会导致解密图像也丢失相应的像素. 和传统的数字全息系统不同, OSH系统可以同时得到两张加密全息图, 如图4所示. 因此, 当一张加密图像的信息丢失严重甚至完全丢失时, 可以对另一张完整的全息图进行数字重建也可以获得解密图像, 如图8所示. 本节实验中, 只对切片1(z1 = 10 mm)进行仿真实验. 图8(a)是余弦加密全息图丢失, 对正弦加密全息图进行解密的结果, 图8(b)是正弦加密全息图丢失, 对余弦加密全息图进行解密的结果, 两个解密结果的相关系数分别为0.6610和0.6506. 虽然单幅机密全息图解密后的噪声较大, 影响视觉效果, 但是仍然可以看到图像的主要特征, 实现对图像信息的识别, 且相关系数较高. 图 8 单幅加密全息图的解密结果 (a) 正弦加密全息图的解密结果; (b) 余弦加密全息图的解密结果 Figure8. Decryption result of single-encrypted hologram: (a) Decryption result of encrypted sine-hologram; (b) decryption result of encrypted cosine-hologram.
对于两幅加密全息图都存在部分信息丢失的情况, 仿真中将图4中的部分像素设置为0, 然后利用两幅加密全息图进行数字重建得到加密结果, 如图9所示. 其中图9(a)和图9(b)是信息丢失25%的正余弦加密图像, 图9(c)是相应的解密结果, 相关系数为0.7418. 图9(d)和图9(e)是信息丢失50%的正余弦加密图像, 图9(f)是相应的解密结果, 相关系数为0.6508. 图9(g)和图9(h)是信息丢失75%的正余弦加密图像, 图9(i)是相应的解密结果, 相关系数为0.5192. 可以看出, 解密后的图像可以识别到图像的主要特征, 可以辨认出图像的轮廓, 具有较好的相关系数, 都在0.5以上, 因此OSH光学加密系统有较强的抗剪裁能力. 图 9 抗剪裁性能模拟结果 (a) 和 (b) 信息丢失25%的正余弦加密图像; (c) 信息丢失25%后的解密结果; (d) 和 (e) 信息丢失50%的正余弦加密图像; (f) 信息丢失50%后的解密结果; (g) 和 (h) 信息丢失75%的正余弦加密图像; (i) 信息丢失75%后的解密结果. Figure9. Simulation results of anti-shear performance: (a) and (b) The sine- and cosine-holograms with 25% occlusion; (c) decryption result with 25% occlusion; (d) and (e) the sine- and cosine-holograms with 50% occlusion; (f) decryption result with 50% occlusion; (g) and (h) the sine- and cosine-holograms with 75% occlusion; (i) decryption result with 75% occlusion.
23.5.系统抗噪声性能分析 -->
3.5.系统抗噪声性能分析
加密后的图像在传输过程中难免会受到噪声的干扰, 因此解密后的图像也会受到一定的影响. 利用不同方差(variance)的高斯噪声和椒盐噪声作为噪声干扰源, 对加密全息图进行叠加干扰, 并通过相关系数来定量评估解密图像的成像效果, 其中所选噪声的均值均为0. 图10(a)—图10(c)分别是受到方差为0.02, 0.05和0.08的高斯噪声(均值为0)干扰后对切片1 (z1 = 10 mm)图像的解密结果, 相对应的相关系数分别为0.6953, 0.6649和0.5787. 图10(d)—图10(f)分别是受到方差为0.02,0.05和0.08的椒盐噪声(均值为0)干扰后对切片1 (z1 = 10 mm)图像的解密结果, 相对应的相关系数分别为0.6825, 0.6034和0.5038. 可以看出, 由于受到噪声的影响, 解密后的图像质量有所下降, 但是解密后图像的相似度均在0.5以上, 解密图像依然可以辨别到明显的边缘特征, 可以识别到原图像的信息. 所以, 本加密系统的抗噪声干扰方面的性能较好, 具有较强的抗噪声能力. 图 10 抗噪声性能模拟结果 (a), (b) 和 (c) 方差为0.02, 0.05和0.08的高斯噪声; (d), (e) 和 (f) 方差为0.02, 0.05和0.08的椒盐噪声 Figure10. Simulation results of anti-noise performance: (a), (b) and (c) Gaussian noise with variance of 0.02, 0.05 and 0.08; (d), (e) and (f) salt and pepper noise with variance of 0.02, 0.05 and 0.08.