Research On Image Encryption Method Based On Dynamic Probability Events

Chaotic based image encryptions the hotspot of image security, but there still exist many deficiencies in the existed chaotic image encryption methods:?The loose coupling of permutation and confusion cannot guarantee the safety of encrypted image; ?The encrypted image's plaintext property is unrelated to the encryption key and encryption procedure so that different encrypted image may experience the same encryption procedure which make it difficult to resist the known or chosen-plaintext attack;?It usually needs more encryption times to achieve a good encryption effect. In selective image encryption method, it usually does not require the whole image is encrypted, and it needs to provide some features of the original image to the potential users with more experience. The kept features can be used for image classification, retrieval, recognition and so on. In traditional selective image encryption methods, there still exist a lot of problems:?Bit plane based selective encryption method has limit ability in adjusting image visual quality;?For threshold based image encryption, if thresholds is not guaranteed correct before and after encryption, it will lead to incorrect image recovery, and to store threshold will bring the additional storage cost; ?In frequency based selective encryption, the amplitude and position changes of frequency coefficients are likely to lead to the spatial pixel overflows, which will affect the visual quality of recovered image. To solve the above problem, the finished research works are followings:(1)In chaotic based image encryption method, there exist many problems such as the loose coupling of permutation and confusion, the encrypted image's plaintext property unrelated to the encryption key and encryption procedure and usually needing more encryption times to achieve a good encryption effect. To address these problems, an image encryption algorithm combined probability interval partitioning with dynamic probability events was proposed. The key parameters and image's SHA-1 value were mapped to different probability ranges and then used to encrypted randomly selected pixels where the fallen probability interval determined the performed encryption event. The proposed method used SHA-1 remapping mechanism to add random disturbance to probability intervals dividing, random selected pixels to be encrypted and encryption procedures. The encryption performance was improved by the global encryption procedure which was related with the user key and newly updated SHA-1 value. By compared with the existing methods, the proposed method coupled permutation with confusion procedure tightly, and different plaintext images led to different encryption procedure. It can use less encryption times to obtained more balanced encryption result. Theory and experiment results showed that the proposed method has a large key space and better encryption performance. It can effectively resist the known or chosen-plaintext attack and enhances the encryption efficiency.(2)In traditional selective image encryption methods, there still exist a lot of problems such as bit plane based selective encryption method has limit ability in adjusting image visual quality; for threshold based image encryption, if thresholds is not guaranteed correct before and after encryption, it will lead to incorrect image recovery, and to store threshold will bring the additional storage cost and in frequency based selective encryption, the amplitude and position changes of frequency coefficients are likely to lead to the spatial pixel overflows, which will affect the visual quality of recovered image. To address these problems, a selective image encryption algorithm combined dynamic encryption events with spatial frequency mixing was proposed. The proposed method mapped the user key and the MD5 and SHA-1 values to different probability intervals and then used the intermediate key to randomly select overlapping pixel block to be encrypted by the corresponding spatial probability frequency encryption events according to the fallen intervals. The proposed method compounded MD5 and SHA-1 together to improve the anti-collision ability of single hash function and generated probability intervals, judging probability, pixel block selection, probability event determining and transform parameters by intermediate key parameters coupled with the user key. The designed encryption included spatial frequency permutation methods based on SCAN language to improve encryption efficiency and avoid frequency coefficient overflowing to reduce influence visual quality of restored image. The proposed method updated both image encryption features and user key to make encryption variables drastically different in different encryption times and also related with the user key and it can also use preset threshold to control image quality and overflow selection. Experiment results showed that the proposed method can effectively resist the known or chosen-plaintext attack, the encrypted image may retain plaintext image approximate which can be used for identification and recognition. The visual aualitv of image restoration can be ensured via threshold judging and the proposed method is extremely sensitive to the encryption times.(3)Combined with the finished research work, an image encryption software based on JAVA was designed, the main classes involved in this software were introduced, and some running results were demonstrated.