Author Affiliations
1College of Software, Liaoning Technical University, Huludao, Liaoning, 125105, China2College of Information Science and Engineering, Yanshan University, Qinhuangdao, Hebei, 0 66004, Chinashow less
Fig. 1. Comparison of feature points before and after screening. (a) Feature point overlap; (b) correction results with large error; (c) extraction results of error correction; (d) filtered feature point matching; (e) corrected image; (f) correctly corrected extraction results
Fig. 2. Comparison of encryption image and original image under different keys. (a) (b) (e) Encryption results with the same key and different initial offset values; (c) (d) encryption results with different keys; (f) original image
Fig. 3. Halftone decryption results. (a)(e) Encryption and decryption images of R channel; (b)(f) encryption and decryption images of G channel; (c)(g) encryption and decryption images of B channel; (d)(h) encryption and decryption images after channel fusion
Fig. 4. Extraction results under different key conditions. (a) Sub-block segmentation key has error in R channel; (b) sub-block segmentation key has error in R and G channels; (c) sub-block segmentation key has error in R, B and G channels; (d) initial key has error in R channel; (e) initial key has error in R and G channels; (f) initial key has error in R, B and G channels; (g) sub-block segmentation key has error in R channel and initial key has error in R and G channels; (h) sub-block segmentation key has error in R and G channels and initial key has error in R channel; (i) sub-block segmentation key and initial key has in R, B and G channels
Fig. 5. Zero watermark generation and copyright logo extraction
Fig. 6. Host images and logo images. (a) Baboon; (b) car; (c) Lena; (d) pepper; (e) plane; (f) sailboat; (g) logo
Fig. 7. Lena graph and detection results under non geometric attacks. (a) Gaussian noise (0.05); (b) salt-and-peppers noise (0.05); (c) median filtering (7×7); (d) JPEG compression(10%)
Fig. 8. Geometric correction results and copyright logo. (a)(h) rotate 25°; (b)(i) rotate 50°; (c)(j) rotate 75°; (d)(k) rotate 100°; (e)(l) rotate 125°; (f)(m) rotate 150°; (g)(n) rotate 175°
Fig. 9. Comparative experiment results. (a) Gaussian filtering; (b) salt-and-peppers noise; (c) JPEG compression; (d) median filtering; (e) Gaussian filtering; (f) rotation attack; (g) translation attack; (h) scaling attack
Logo image | False alarm rate of input image |
---|
Baboon | Car | Lena | Pepper | Plane | Sailboat |
---|
Baboon | 1.0000 | 0.5602 | 0.5640 | 0.5672 | 0.5925 | 0.5558 | Car | 0.5585 | 1.0000 | 0.5731 | 0.5580 | 0.4810 | 0.5585 | Lena | 0.5697 | 0.5637 | 1.0000 | 0.5671 | 0.5729 | 0.5604 | Pepper | 0.5605 | 0.5588 | 0.5636 | 1.0000 | 0.5329 | 0.5605 | Plane | 0.5650 | 0.6008 | 0.5669 | 0.5537 | 1.0000 | 0.5909 | Sailboat | 0.5885 | 0.5526 | 0.5573 | 0.5857 | 0.5324 | 1.0000 |
|
Table 1. Experimental results of false alarm rate
Attack type | Parameter | Baboon | Car | Lena | Pepper | Plane | Sailboat |
---|
Gaussian noise | 0.02 | 0.9887 | 0.9877 | 0.9860 | 0.9900 | 0.9906 | 0.9948 | 0.05 | 0.9820 | 0.9801 | 0.9780 | 0.9833 | 0.9849 | 0.9925 | Salt &pepperNoise | 0.02 | 0.9935 | 0.9936 | 0.9924 | 0.9946 | 0.9945 | 0.9974 | 0.05 | 0.9896 | 0.9888 | 0.9890 | 0.9923 | 0.9912 | 0.9956 | Speckle noise | 0.02 | 0.9939 | 0.9914 | 0.9931 | 0.9957 | 0.9926 | 0.9975 | 0.05 | 0.9883 | 0.9851 | 0.9881 | 0.9933 | 0.9889 | 0.9952 | Median filtering | 3×3 | 0.9949 | 0.9964 | 0.9982 | 0.9981 | 0.9975 | 0.9984 | 7×7 | 0.9890 | 0.9897 | 0.9930 | 0.9955 | 0.9905 | 0.9947 | Wiener filtering | 3×3 | 0.9986 | 0.9988 | 0.9992 | 0.9992 | 0.9990 | 0.9994 | 7×7 | 0.9953 | 0.9947 | 0.9965 | 0.9969 | 0.9964 | 0.9978 | Low pass filtering | 3×3 | 0.9984 | 0.9986 | 0.9982 | 0.9985 | 0.9983 | 0.9991 | 7×7 | 0.9976 | 0.9977 | 0.9975 | 0.9980 | 0.9974 | 0.9986 | JPEG compression | 10 | 0.9810 | 0.9800 | 0.9757 | 0.9844 | 0.9832 | 0.9910 | 50 | 0.9955 | 0.9947 | 0.9947 | 0.9954 | 0.9960 | 0.9978 |
|
Table 2. Experimental results of non geometric attack (NC)
Attack type | Parameter | Baboon | Car | Lena | Pepper | Plane | Sailboat |
---|
Rotation | 0° | 1.0000 | 1.0000 | 1.0000 | 1.0000 | 1.0000 | 1.0000 | 25° | 0.8826 | 0.8077 | 0.8667 | 0.8717 | 0.8237 | 0.9359 | 50° | 0.8477 | 0.7679 | 0.8489 | 0.8568 | 0.8117 | 0.9170 | 75° | 0.9091 | 0.8608 | 0.9088 | 0.9148 | 0.8326 | 0.9464 | 100° | 0.9435 | 0.9044 | 0.9336 | 0.9361 | 0.9212 | 0.9606 | 125° | 0.8615 | 0.7780 | 0.8469 | 0.8556 | 0.8103 | 0.9220 | 150° | 0.8167 | 0.7876 | 0.8566 | 0.8716 | 0.8235 | 0.9199 | 175° | 0.9596 | 0.9515 | 0.9725 | 0.9725 | 0.9599 | 0.9845 | Move | 5 lines and 5 columns | 0.9937 | 0.9899 | 0.9906 | 0.9924 | 0.9823 | 0.9961 | 10 lines and 20 columns | 0.9793 | 0.9643 | 0.9755 | 0.9789 | 0.9644 | 0.9875 | Scaling | 0.25 | 0.9945 | 0.9943 | 0.9974 | 0.9972 | 0.9912 | 0.9974 | 1.25 | 0.9997 | 0.9998 | 0.9999 | 0.9998 | 0.9991 | 0.9999 |
|
Table 3. Experimental results of geometric attack (NC)
Step | Time /s |
---|
Copyright watermark preprocessing | 0.1821 | Generate zero watermark | 0.5643 | Original image | 0.6009 | 20° rotation | 0.6026 | 45° rotation | 0.6241 | Salt-and-pepper noise 0.08 | 0.6344 | Wiener filter 9×9 | 0.6089 |
|
Table 4. Run time test table