不连续四元值神经网络的限时同步及在医学图像保护中的应用

打开文本图片集

中图分类号：TP391.4 文献标识码：A 文章编号：2096-4706（2026）05-0136-06

Fixed/Prescribed-time Synchronization of Discontinuous Quaternion-valued Neural Networks with Applications to Medical Image Protection

WANG Gang，QINXuejiao，NIUMengqin， ZHANGWenbin，WANGXingwei （XinjiangSecondMedicalCollege，Karamay834oo0，China）

Abstract：Toaddress theneedforsecuringhigh-dimensionalmedicalimagedata，thispaperinvestigates thefixed/prescribedtime synchronizationof discontiuousQuaterion-Valued Neural Networks （QVNNs）and aplies ittomedical image encryption. A discontinuousQVNN model isconstructed，andbyintegrating the non-separationmethodand differential inclusiontheory， twotypesofcontrolstrategiesaredesignedtoacheveapidsynchronzation.Theoreticalanalysisprovidessuffcientconditions forsynchronization，supportedbynumerical simulations thatverifytheeffctivenessofthecontrolstrategies.Furtherore，a chaos synchronization-based image encryptionscheme is proposed， which protectsmedical images through pixel permutation anddiffusion.Experimentalresults demonstrate thattheproposed schemeachieves excelent encryption performance in termsof informationetropystogamdistrution，ndoelationcents，ectivelyesurngthescuityofmedicalges.

Keywords：Quaterion-valued NeuralNetwork; discontinuousactivation function;fixed/prescribed-time; synchronization; medical image protection

0 引言

在当今数字化时代，神经网络凭借其在模式识别[、图像处理[2]等诸多领域的应用价值，受到了广泛关注。（剩余11018字）