Research On Anti-distortion Holographic Digital Watermarking Algorithm Based On Image Feature Block

With the rapid development of computer network communications,there will be a variety of serious problems that harm social interests without a strict network supervision system.Digital watermarking technology has gradually become an effective means of information anti-counterfeiting,but the current mature watermarking algorithms are not enough to resist distortion attacks.In order to make the watermark resistant to distortion attacks,this thesis proposes an anti-distortion holographic digital watermarking algorithm based on image feature blocks.First,the carrier image is divided into different sub-image feature blocks according to the image characteristics,the SIFT feature points of each sub-image feature block are extracted,and the carrier feature blocks are filtered according to the distribution position and stability of the feature points,finally all image features suitable for embedding watermarks are selected Piece.The original watermark image is transformed by Fresnel diffraction to generate four holographic watermark images,and I1 is extracted from them.After a certain filtering rule,the appropriate feature points are selected,and the surrounding 8×8 area is selected with the center as the center.Wavelet transform is performed on the area to extract the Then,the holographic watermark image is divided into blocks as required and embedded according to the rules;according to the same embedding method,the holographic watermark is embedded in each selected image feature block,and the whole image is finally synthesized.When extracting the watermark,extract the watermark image in each sub-image feature block separately and evaluate the quality,and compare and select the watermark image with the better quality.The programming has realized the embedding and extraction part of the watermark and the robustness test part.The experimental results show that the image quality of the embedded watermark using this algorithm has no obvious change,and the extracted watermark image is clearer.Since the distribution area of the watermark image(image block)is enlarged while the stability of the embedded points(feature points)is improved,the watermark has strong robustness and the algorithm is feasible.In order to successfully extract the watermark from the watermarked image after distortion,this thesis proposes a watermark correction algorithm based on image feature blocks and feature points.The correction of the entire image is regarded as the correction of the partial image.This article describes two correction methods: first one is to divide the distorted image and the original image into blocks according to their image features,extract the feature points in the feature blocks,and then match the feature points between the image feature blocks,according to the matching feature points Calculate the amount of change based on the slope of change,and finally correct other pixels based on the amount of change,but this method is only suitable for regular distortion(that is,the amount of change of pixels in each image feature block is the same),there are certain limitations;the other one is to divide the image into image feature blocks and then extract the feature points and match them,take the image feature block as a unit,divide the triangle according to the feature points,and correct the triangle and other pixels according to the coordinate relationship of the triangle vertices.The triangles divided by this method are thin and the triangles themselves are stable,so there are almost no limitations.Therefore,this thesis finally adopts the latter correction method.After the correction is completed,the watermark information in the characteristic block is extracted,the watermark quality is evaluated and the selection is excellent,and finally the distortion attack test is performed.Experiments show that the correction algorithm in this thesis can successfully extract a clear watermark image after the image is subjected to a certain distortion attack,which proves that the algorithm has strong anti-distortion ability.