A METHOD FOR EMBEDDING AND EXTRACTING DIGITAL WATERMARK WITHOUT DISTORTING WATERMARKED IMAGE USING A COMPLEX VALUED NEURAL NETWORK

摘要

THE PRESENT INVENTION RELATES TO A METHOD FOR EMBEDDING AND EXTRACTING DIGITAL WATERMARK WITHOUT DISTORTING WATERMARKED IMAGE USING A COMPLEX VALUED NEURAL NETWORK, CHARACTERIZED BY THE STEPS OF: CONVERTING A WATERMARK INTO BINARY TARGET DATA; THEN DECOMPOSING A HOST IMAGE AND SELECTING A PLURALITY OF BLOCKS FROM THE HOST IMAGE FOR WATERMARKING; THEN TRANSFORMING EACH SELECTED BLOCK FROM SPATIAL DOMAIN TO FREQUENCY DOMAIN USING FAST FOURIER TRANSFORM ALGORITHM; THEN SORTING EACH TRANSFORMED BLOCK INTO A WATERMARK BIT TIME LINE USING A POSITION SORTER; THEN FEEDING THE COMPLEX INPUT DATA AS TRAINING DATA INTO THE COMPLEX VALUED NEURAL NETWORK WHICH COMPRISES OF A COMPLEX VALUED FEED FORWARD NEURAL NETWORK AND A COMPLEX BACK-PROPAGATION ALGORITHM FOR PROCESSING; THEN CLASSIFYING THE NETWORK OUTPUT INTO A SAFE REGION AND AN ATTACK VULNERABLE REGION; THEN MAPPING THE BINARY TARGET DATA OF THE WATERMARK TO THE SAFE REGION. DRAWING ACCOMPANYING ABSTRACT: