CN-116416112-B - Medical image self-recovery method based on image restoration algorithm

CN116416112BCN 116416112 BCN116416112 BCN 116416112BCN-116416112-B

Abstract

The invention discloses a medical image self-recovery method based on an image restoration algorithm, which is characterized in that disclosed medical image samples are collected and arranged, a low-frequency sub-band LL3 is selected as an approximate compression version of an original image after DWT is carried out on the medical image, and a watermark value is generated. And extracting areas such as contours, edges and the like of the medical image by using the Hessian matrix, and embedding watermark values into the areas such as the contours, the edges and the like by using the LSB algorithm to generate a watermark image. When being attacked, the method extracts a reference watermark from the attack image, detects a tampered area of the watermark image by using the extracted watermark value, and restores the tampered area by using similar information of the attack image by means of an image restoration algorithm. The method can well protect the information of the original medical image by embedding the watermark value in the characteristic region of the original medical image. When the tampered area is restored, the tampered area can be restored by utilizing the similar information of the medical image, so that the method has practical significance and achieves a good effect.

Inventors

BI XIULI
LI MINGYUE
XIAO BIN
LIU BO

Assignees

重庆邮电大学

Dates

Publication Date: 20260512
Application Date: 20230316

Claims (6)

1. A medical image self-restoration method based on an image restoration algorithm, comprising the steps of: generating a watermark value by taking the medical image as an original image; extracting characteristic information of an original image by using a Hessian matrix; The LSB algorithm is utilized to embed the generated watermark value in the original image to generate a watermark image embedded with the watermark, and the method specifically comprises the following steps: for the first watermark value, it is assumed that the watermarked pixel value is equal to the original pixel value If the sequence of embedding the watermark is greater than 1 and the current embedded watermark value is a positive integer, the embedding formula is as follows: after embedding the watermark value, the marking bit of the pixel is set to 1 as follows: If the watermark value embedded at present is a negative integer, taking absolute value, embedding the watermark value by utilizing the LSB algorithm, wherein the embedding formula is as follows: Wherein, the Represents the first The sequence number of the individual watermark values, Representing the first image of the original image Line 1 The number of columns in a row, Representing the pixel values after embedding the watermark I refers to the original image, Refers to the ith watermark value; Traversing watermark values one by one Until all pixel values are embedded in the original image to obtain a watermark image embedded with the watermark; When the watermark image embedded with the watermark is attacked, extracting a reference watermark from the attack image, and detecting a tampered area by using the extracted watermark value; After the tampered area is detected, the tampered area is restored by utilizing the similar information of the attack image by means of an image restoration algorithm, wherein the restoration of the tampered area by utilizing the similar information of the attack image by means of the image restoration algorithm comprises the steps of finding the similar information of the tampered area in the attack image by means of the image restoration algorithm, and copying the similar information of the attack image to be pasted on the tampered area to restore the tampered area.
2. The method for self-restoring a medical image based on an image restoration algorithm according to claim 1, wherein generating a watermark value using the medical image as an original image includes selecting a low frequency subband LL3 as an approximately compressed version of the original image after DWT of the original image, and performing quantization encoding on coefficients of the LL3 subband and generating the watermark value.
3. The image restoration algorithm-based medical image self-restoration method according to claim 2, wherein the method of quantitatively encoding the coefficients of the LL3 subband is differential pulse encoding.
4. The image restoration algorithm-based medical image self-restoration method according to claim 1, wherein the detection of the tampered region using the extracted watermark value comprises the steps of: Extracting watermark values from the attack image; obtaining an approximately compressed version of an original image using differential pulse encoded watermark values ; Comparing the attack image with the obtained To obtain an approximate tampered region ; Tampered areas to be obtained Thresholding and morphological processing to obtain the final effective tamper range Wherein T refers to an attack image; refers to a tampered region detected after a thresholding operation.
5. The image restoration algorithm-based medical image self-restoration method according to claim 1, wherein the restoration of the tampered region by means of the image restoration algorithm using similar information of the attack image itself comprises the steps of: If the detected tampered pixel is tampered, determining a tampered block by taking the current tampered pixel as a center; Wherein, the Is an attacked image after embedding the watermark; In order to detect the tampered region, Is a known part; in known areas of the extracted watermark image Searching the best matching block, and searching the matching block The following are provided: Wherein, the Belonging to a known region And the value range is related to the size of the attack area; searching the best matching block of the tampered block in all the matching blocks The formula is as follows: Wherein, the Is a tampered block centered around tampered pixel x, Is the pixel that is currently being matched, Is a matching block centered on the matching pixel p in the known region from which the watermark was extracted, and the matching block is measured by Euclidean distance Tamper block Is a degree of similarity of (2); The process of mapping the spatial locations of similar blocks to known areas of the attack image is as follows: Wherein, the Similar information for known regions of the attack image; And obtaining a similar block of the tampered block by using an image restoration algorithm, copying information of the similar block, and pasting the information in a tampered area to obtain a tampered restored image.
6. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor implements the steps of the image restoration algorithm based medical image self-restoration method according to any of claims 1 to 5.

Description

Medical image self-recovery method based on image restoration algorithm Technical Field The present disclosure relates to the technical fields of digital image processing, computer vision, signal processing, etc., and in particular to a medical image self-recovery method based on an image restoration algorithm. Background The statements in this section merely provide background information related to the present disclosure and may constitute prior art. In carrying out the present invention, the inventors have found that at least the following problems exist in the prior art. In recent years, image processing technology is rapidly developed, various complex image modification software is increased, and people can modify and spread digital multimedia content according to own ideas by using the software. Therefore, the informatization brings convenience to the work and life of people and also causes some potential safety hazards. The attacker uses a powerful multimedia editing tool to maliciously tamper and forge the content of the digital multimedia information, so that the naked eyes cannot distinguish the authenticity, and the events of copyright infringement and malicious tampering are increased continuously. Digital images are widely used for information delivery as important carriers of multimedia information. How to use digital watermarking technology to prevent information leakage and protect the safety of multimedia information is a hot topic of people. Digital watermarking has been widely used in various digital multimedia, and its carrier may be text, image, video, music, etc. The main idea of digital watermarking is to embed secret information in the form of a watermark into the multimedia information itself to create a protective carrier, based on the redundancy of the carrier, and the embedded secret information is invisible. People cannot perceive it only by the visual or auditory system. The authenticity and integrity of the digital multimedia is checked by means of information extracted from the protection carrier. The digital watermark is mainly divided into two types according to the application, wherein the first type is represented by a robust watermark and is mainly used for protecting copyright, such as author information, publishing information and the like. The second category is represented by fragile watermarks, which are mainly used for judging whether the content is tampered with. Initially, fragile watermarking techniques only embedded an authentication watermark in the original image to detect the authenticity of the content. The authentication watermark is typically generated from a hash value of the original image, used to detect the tampered area, but cannot recover the tampered area. In order to realize recovery of the tampered area while detecting the tampered area, most of the existing fragile watermarking methods use a dual watermarking mechanism (authentication watermark and reference watermark) to embed watermark information into an original image to generate a protection image. In the event that the protected image is corrupted, a reference watermark message may be extracted to recover the tampered image. The reference watermark is typically generated by means of low frequency coefficients of each block in the DCT, low frequency coefficients of the wavelet transform, and a halftone version of the original image, representing the main content of the original image, for recovering the tampered region. A common way to achieve restoration is a self-embedding watermarking algorithm, e.g. embedding a highly compressed version of the image in the image itself or restoring the data with the basic information of the original image. Such techniques, which have both localization and restoration of tampered areas, have received considerable attention from researchers. The prior tamper detection and self-recovery method is used for embedding watermark information into an original image, not only can a tamper area be detected, but also the tamper area can be recovered by utilizing information hidden in other areas of the image, but also has some defects in the field of medical images (1) because the medical image is a sensitive image, a large amount of embedded watermark information is likely to destroy the subjective effect of the original image, (2) when the tamper area is detected, the tamper area is approximately recovered by utilizing a reference watermark extracted from an attack image to fill the tamper area, so that the content of the tamper area is ignored, and (3) the medical image is different from other images, and has much similar information, but the existing method ignores the similar information of the medical image and does not utilize the similar information of the medical image. Disclosure of Invention In view of the above, it is an object of the present invention to solve some of the problems of the prior art, or at least to alleviate them. A medical image self-re