Bharath V Nair, Vismaya V S, Sishu Shankar Muni, Ali Durdu
{"title":"深度学习与混沌:图像加密和解密的组合方法","authors":"Bharath V Nair, Vismaya V S, Sishu Shankar Muni, Ali Durdu","doi":"arxiv-2406.16792","DOIUrl":null,"url":null,"abstract":"In this paper, we introduce a novel image encryption and decryption algorithm\nusing hyperchaotic signals from the novel 3D hyperchaotic map, 2D memristor\nmap, Convolutional Neural Network (CNN), and key sensitivity analysis to\nachieve robust security and high efficiency. The encryption starts with the\nscrambling of gray images by using a 3D hyperchaotic map to yield complex\nsequences under disruption of pixel values; the robustness of this original\nencryption is further reinforced by employing a CNN to learn the intricate\npatterns and add the safety layer. The robustness of the encryption algorithm\nis shown by key sensitivity analysis, i.e., the average sensitivity of the\nalgorithm to key elements. The other factors and systems of unauthorized\ndecryption, even with slight variations in the keys, can alter the decryption\nprocedure, resulting in the ineffective recreation of the decrypted image.\nStatistical analysis includes entropy analysis, correlation analysis, histogram\nanalysis, and other security analyses like anomaly detection, all of which\nconfirm the high security and effectiveness of the proposed encryption method.\nTesting of the algorithm under various noisy conditions is carried out to test\nrobustness against Gaussian noise. Metrics for differential analysis, such as\nthe NPCR (Number of Pixel Change Rate)and UACI (Unified Average Change\nIntensity), are also used to determine the strength of encryption. At the same\ntime, the empirical validation was performed on several test images, which\nshowed that the proposed encryption techniques have practical applicability and\nare robust to noise. Simulation results and comparative analyses illustrate\nthat our encryption scheme possesses excellent visual security, decryption\nquality, and computational efficiency, and thus, it is efficient for secure\nimage transmission and storage in big data applications.","PeriodicalId":501167,"journal":{"name":"arXiv - PHYS - Chaotic Dynamics","volume":"38 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Deep Learning and Chaos: A combined Approach To Image Encryption and Decryption\",\"authors\":\"Bharath V Nair, Vismaya V S, Sishu Shankar Muni, Ali Durdu\",\"doi\":\"arxiv-2406.16792\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In this paper, we introduce a novel image encryption and decryption algorithm\\nusing hyperchaotic signals from the novel 3D hyperchaotic map, 2D memristor\\nmap, Convolutional Neural Network (CNN), and key sensitivity analysis to\\nachieve robust security and high efficiency. The encryption starts with the\\nscrambling of gray images by using a 3D hyperchaotic map to yield complex\\nsequences under disruption of pixel values; the robustness of this original\\nencryption is further reinforced by employing a CNN to learn the intricate\\npatterns and add the safety layer. The robustness of the encryption algorithm\\nis shown by key sensitivity analysis, i.e., the average sensitivity of the\\nalgorithm to key elements. The other factors and systems of unauthorized\\ndecryption, even with slight variations in the keys, can alter the decryption\\nprocedure, resulting in the ineffective recreation of the decrypted image.\\nStatistical analysis includes entropy analysis, correlation analysis, histogram\\nanalysis, and other security analyses like anomaly detection, all of which\\nconfirm the high security and effectiveness of the proposed encryption method.\\nTesting of the algorithm under various noisy conditions is carried out to test\\nrobustness against Gaussian noise. Metrics for differential analysis, such as\\nthe NPCR (Number of Pixel Change Rate)and UACI (Unified Average Change\\nIntensity), are also used to determine the strength of encryption. At the same\\ntime, the empirical validation was performed on several test images, which\\nshowed that the proposed encryption techniques have practical applicability and\\nare robust to noise. Simulation results and comparative analyses illustrate\\nthat our encryption scheme possesses excellent visual security, decryption\\nquality, and computational efficiency, and thus, it is efficient for secure\\nimage transmission and storage in big data applications.\",\"PeriodicalId\":501167,\"journal\":{\"name\":\"arXiv - PHYS - Chaotic Dynamics\",\"volume\":\"38 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-06-24\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - PHYS - Chaotic Dynamics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2406.16792\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - PHYS - Chaotic Dynamics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2406.16792","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Deep Learning and Chaos: A combined Approach To Image Encryption and Decryption
In this paper, we introduce a novel image encryption and decryption algorithm
using hyperchaotic signals from the novel 3D hyperchaotic map, 2D memristor
map, Convolutional Neural Network (CNN), and key sensitivity analysis to
achieve robust security and high efficiency. The encryption starts with the
scrambling of gray images by using a 3D hyperchaotic map to yield complex
sequences under disruption of pixel values; the robustness of this original
encryption is further reinforced by employing a CNN to learn the intricate
patterns and add the safety layer. The robustness of the encryption algorithm
is shown by key sensitivity analysis, i.e., the average sensitivity of the
algorithm to key elements. The other factors and systems of unauthorized
decryption, even with slight variations in the keys, can alter the decryption
procedure, resulting in the ineffective recreation of the decrypted image.
Statistical analysis includes entropy analysis, correlation analysis, histogram
analysis, and other security analyses like anomaly detection, all of which
confirm the high security and effectiveness of the proposed encryption method.
Testing of the algorithm under various noisy conditions is carried out to test
robustness against Gaussian noise. Metrics for differential analysis, such as
the NPCR (Number of Pixel Change Rate)and UACI (Unified Average Change
Intensity), are also used to determine the strength of encryption. At the same
time, the empirical validation was performed on several test images, which
showed that the proposed encryption techniques have practical applicability and
are robust to noise. Simulation results and comparative analyses illustrate
that our encryption scheme possesses excellent visual security, decryption
quality, and computational efficiency, and thus, it is efficient for secure
image transmission and storage in big data applications.