Flexible digital image self-recovery method based on fountain coding

Abstract

The invention discloses a flexible digital image self-recovery method based on fountain coding. The method comprises steps of: first classifying image blocks according to texture complexity, obtaining classification data, and assigning different coding lengths and embedding capacity for each classification; secondly, performing discrete cosine transform (DCT) on the images to generate recovery data; subjecting the recovery data and the classification data to fountain coding to generate the reference data, subjecting the reference data and original information to a Hash function to generate authentication data for obtaining the stego image; and finally performing tamper identification according to the authentication data and reconstructing the original image through the recovery data. The method carries out variable-length coding and variable-capacity embedding on the images, and the higher the texture complexity is, the longer the coding length is so that the recovery quality of the whole image can be ensured. Non-uniform embedding method ensures the quality of the stego image. The reference data is embedded into the whole image by fountain coding so as to ensure that the image has a large maximum recoverable ratio.

Description

technical field [0001] The invention relates to a digital image self-recovery method for image authenticity and integrity authentication, and its basic feature is to carry out variable-length coding and variable-capacity embedding on the image, and apply fountain coding to the reference data to embed into the whole image. Background technique [0002] Digital image self-restoration technology, as an important branch of image authentication technology, is an effective means of image protection. It embeds the original image's own information as reference information into the original image. Extracting its own information can verify the authenticity of the image content and approximate the content of the original image. Carrying out this research is of great significance for ensuring public trust in order, maintaining judicial justice, and combating crimes. [0003] At present, common self-recovery methods mostly use the least significant bit (LSB) bit-plane replacement method...

AI Technical Summary

Problems solved by technology

[0008] However, these methods usually use uniform encoding and uniform embedding. When the embedding capacity is low, many details will be lost when the tampered image is restored, the quality of the reconstructed image is low, and the maximum recoverable ratio is also small, which cannot be used in practical applications. Requirements; when the embedding capacity increases, the reconstructed image quality and the maximum recoverable ratio are improved, but the quality of the dense image decreases significantly, and false contours may occur

Images