Genetic Based Medical image Steganography using Frequency Feature

Abstract

Medical diagnosis highly depends on patient test reports, some of are just value while newlinesome are images. Using medical imagery to diagnose diseases has become increasingly newlineimportant in recent years. Medical image diagnosis may need some experts, so newlineauthenticity of image is required. Because these photographs are transferred over the newlineinternet, they require a high level of security. If the data in these photographs is newlinevulnerable to unlawful use, it could cause serious difficulties. Securing photographs can newlinebe done in a variety of ways. Many of scholars has resolved this issue by image newlinesignature embedding that help to verify the authenticity of the model. This work has newlineproposed a medical image security model SFCHDH (Selected Frequency Coefficient newlineBased Histogram Data Hiding) that hides patient secret information in binary format. newlineWork has transform image into discrete wavelet frequency domain and histogram newlinefeatures were extract from the DWT coefficient values. DWT coefficients were passed newlineinto genetic algorithm that will cluster values into embedding and non-embedding newlinepositions. As per shifting method of histogram embedding of data was done by the newlinemodel. This work has proposed another model that add verification data into image by newlineusing wolf optimization genetic algorithm and histogram shifting. DWT feature were newlineextract from the image for identifying low frequency regions. In order to identify less newlineaffective pixel coefficient wolf optimization algorithm identify such pixel sets. To newlinerecover image at the receiver end histogram shifting approach was used. Experiment newlinewas done on real patient medical images and implementation was done on MATLAB. newlineResults shows that proposed models has reduced the embedding time and maintains the newlineimage quality by evaluating PSNR, SNR values, as compared to existing models. Result newlineshows that PSNR value was improved by 80.36, similarly SNR value was improved by newline40.9 as compared to [69]. In attack environment extraction percentage of proposed newlinemodel was improved by 39.77 under compression attac