KR100459922B1 - Digital image watermarking using region segmentation - Google Patents

Abstract

The present invention relates to a method of embedding a watermark in a digital image using area division in the computer and software fields. In particular, the watermark is inserted by analyzing the semantic importance of the elements (objects) of the original image to be watermarked and segmenting the object area having high semantic importance and encrypting and inserting the watermark into the divided object area. Even if the domains were extracted or copied separately, they could still be claimed. Even if a user (third party) extracts the area and the watermark by anticipating a region having high semantic importance by encrypting and inserting the watermark information, the extracted watermark information can still be protected without being exposed because it is encrypted. .

Accordingly, the present invention comprises the steps of extracting a region having a high semantic importance by performing region division processing from the original image; Encrypting and inserting watermark information into the extracted area; Dividing a partial region of interest to a user (third party) from the watermarked image; Compositing the divided partial regions with an arbitrary background image; A watermark embedding method of a digital image is provided, comprising extracting a watermark from the synthesized image.

Description

Digital image watermarking using region segmentation

The present invention relates to a digital image watermark embedding method using area division in the computer and software fields. More specifically, the present invention relates to a digital watermarking technology that enables a copyright and ownership claim to be made even if a watermarked region is extracted or copied and used separately.

Digital watermarking is a copyright protection solution for digital content that prevents illegal duplication and effectively protects the copyright and ownership of data owners by embedding user ID or personal information into digital content.

Conventional digital image watermarking technology performs processing in the statistical and frequency domain without considering the components (objects) of the image. However, not all object regions constituting an image on an image are necessarily regions of interest that are illegally used or stolen by a third party.

The interests of third parties as intellectual property infringers are centered on the area of some objects that make up an image. When some regions in one image are divided and synthesized into other images, the regions of interest of third parties are regions of high significance, which are targets of illegal use or theft on the original image.

The correlation of the originally inserted watermark from the synthesized image is determined by segmenting the region before inserting the conventional watermarking information to determine the region of high significance and inserting the encrypted watermark. The watermark can still be extracted even if it is divided and used, and since the watermark information is encrypted, the inserted watermark information can be protected without being exposed.

(1) W. Bender, D. Gruhl, N. Morimoto, and A. Lu, IBM Systems Journal, Vol. 35, pp. 313-336, 1996, [2] ML Miller, IJ Cox, JMG Linnartz, T. Kalker, Digital Signal Processing in Multimedia Systems, KK Parhi, T. Nishitani (eds.), Marcell Dekker, Inc., pp. 461-485, NY, 1999, [3] C. Lin, ML Miller, YM Liu, Proc. Of SPIE, vol. 3971, Security and Watermarking of Multimedia Contents, 2000) categorize watermarking technologies into two main categories according to how and how to insert them.

One method is to insert watermark information into a sample of data, and the other is to insert frequency components. Based on these methods, we describe algorithms that can take advantage of each method and make up for the disadvantages. References make general and detailed references to worker marking techniques.

The watermarking technology, which is mainly mentioned in these documents, has the advantage that it is useful for multimedia, but considering the various attacks and stability to remove the workmark, the contents of the data, that is, the components (objects) Sufficient semantic importance should be considered.

Accordingly, the present invention is to solve the above problems, the present invention is to analyze the semantic importance of the components (objects) of the original image that is the watermark insertion object to segment and segment the object area of high semantic importance By inserting an encrypted watermark in the object area, the object can still claim copyright and ownership even if the watermarked area is extracted or copied and used separately.

As a technical idea for achieving the above object of the present invention

Area segmentation is performed using features of image data including shape, meaning, color, saturation, visual characteristics of brightness, directionality and movement of an object from an original image. Extracting a facial region using facial organ position information including skin color or pupil position from the image;

Inserting an encrypted watermark into the extracted area; And

Extracting the originally inserted watermark from the damaged or synthesized image data including the image data into which the watermark information is inserted.

Provided is a watermarking embedding method of a digital image comprising a.

1 is a basic conceptual diagram of a digital image watermarking method using region division according to the present invention.

2 is a photograph showing an arbitrary face image as an example of a watermark embedding target.

FIG. 3 is a photograph showing an image showing a face region (object region) divided as a region having high semantic importance.

4 is a photograph showing an image visually showing the encrypted watermark information and the insertion position.

5 is a photograph showing an image in which an encrypted watermark is invisibly inserted.

6 is a photograph showing an arbitrary background image prepared for synthesis.

FIG. 7 is a photograph showing an image obtained by synthesizing an object region divided from a face image into a background image.

8 is a photograph showing an image showing the position and the watermark information detected and decoded from the synthesized image.

Hereinafter, with reference to the accompanying drawings, the configuration and operation of the embodiment of the present invention will be described in detail.

In the present invention, a face picture, an image such as an ID card picture, is set as an example of the watermark insertion target image, and its configuration and operation will be described. In a face photograph or image, a face region and a region of a facial organ (a forehead, an eye, an eyebrow, or a mouth) occupy a high significance area in a photograph or an image.

FIG. 1 is a basic conceptual diagram of a digital image watermarking method using area segmentation according to the present invention. First, an area segmentation process (2) is performed from an original image (1) to extract an area having high semantic importance and is encrypted in the area. Watermark information is inserted (3). The watermark is still separated from the watermarked image 3 by dividing (4) some areas of interest to the user (third party) and combining them with another image (background image) 5 without permission. It can be seen that can be extracted completely (7).

FIG. 2 illustrates an arbitrary face image that is a watermarking insertion target and detects a face region using information such as skin color or eye position from an arbitrary face image. 3 shows an image obtained by detecting and segmenting a face region.

As described above, after finding the area and the position of the face area or organ, these areas are divided from other backgrounds (object areas). Facial areas and organs are highly relevant because they are deeply involved in determining human impressions.

After segmentation, watermark 10 information is inserted as shown in FIG. 4 visually shows what information the watermark 10 is inserted at.

In FIG. 5, the watermark 10 (e.g., KIST mark) is inserted and extracted by a watermarking method considering invisibility and encryption, so that even if a part of the entire image is damaged or a part of the divided area is copied and used without permission, Copyright and ownership can still be claimed, and the watermark is encrypted so that the watermark information can be protected without being exposed even if an area with high semantic importance is extracted by the user (third party).

FIG. 6 shows an arbitrary background image used for compositing, and FIG. 7 shows an example of an image synthesized using a region divided from the original image. 8 shows that the inserted watermark 10 can still be extracted even if the divided area is copied and used without permission.

As described above, according to the method of inserting a watermark of a digital image using the area segmentation according to the present invention, an image that does not exist in the existing watermarking method by segmenting and inserting a watermark by region-partitioning a region of interest having a high significance significance. Copyright protection of component (object) areas can now be achieved.

In addition, by protecting not only copyright protection but also content information included in the image data as a whole, it may cause a creative development motivation for creators and developers. This will ultimately bring profits to developers by increasing the motivation for creative development, further contributing to the protection of the content industry and the development of a stagnant software industry.

Claims (4)

Area segmentation is performed using features of image data including shape, meaning, color, saturation, visual characteristics of brightness, directionality and movement of an object from an original image. Extracting a facial region using facial organ position information including skin color or pupil position from the image; Inserting an encrypted watermark into the extracted area; And Extracting the originally inserted watermark from the damaged or synthesized image data including the image data into which the watermark information is inserted. Watermark embedding method of a digital image comprising a. delete The method according to claim 1, wherein the encrypted watermark is inserted into an area of high semantic importance of interest to a user (third party), so that part of the entire image is damaged or partly divided areas are copied and used without permission. The method of inserting a watermark of a digital image, characterized in that it is possible to claim copyright and ownership. The method of claim 1, wherein the watermark is encrypted so that watermark information can be protected without being exposed even if a region of high significance is extracted by a user (third party). Way.