KR100912125B1 - System and method for evaluating and certifying image identifier - Google Patents

Abstract

The present invention relates to a still image identification software evaluation / authentication system and method.

The present invention evaluates a software product for identifying a deformed image according to the copying or intentional attack of a still image or the type of the Internet or a playback device, and automatically generates a certificate according to the evaluation result so that the user can recognize the still image. It is about evaluation / certification of still image identification software that allows you to select the right product for your purpose based on the certificate.

Still image, evaluation, identification, evaluation environment

Description

Still image identification software evaluation / certification system and method {SYSTEM AND METHOD FOR EVALUATING AND CERTIFYING IMAGE IDENTIFIER}

1 is a view showing the configuration of a still image identification software evaluation / authentication system according to the present invention,

2 is a diagram showing an embodiment of a still image identification software evaluation / authentication system according to the present invention;

3 is a flowchart illustrating a still image identification software evaluation / authentication method according to the present invention.

Description of the main parts of the drawing

110: receiving module 120: still image DB

130: evaluation configuration module 140: conversion image DB

150: identification test module 160: still image identification software

170: evaluation module 180: certificate generation module

The present invention relates to evaluation and certification of still image identification software, and more particularly, to evaluate a software product for identifying a modified image according to the copying or intentional attack of a still image or a type of an Internet or a playback device, and accordingly The present invention relates to a still image identification software evaluation / authentication system and method for setting a standard of an evaluation system and a certificate for certification so that a user can select a required software product according to his purpose based on the contents of a certificate of a still image identification technology. .

As demand for digital content increases, a vast amount of multimedia content (video, music, video, etc.) is constantly being created, produced, distributed, and serviced. Among them, digital still images (hereinafter referred to as "still images") are exploding in popularity with the proliferation of high performance, portable digital cameras, mass storage devices and portable storage media, and price drops. Accordingly, there is an increasing demand for a product that efficiently locates such still images as needed and accurately identifies the desired objects from others. This technique is called still image retrieval technology.

Until now, a technique for retrieving still images has been developed and used a lot of methods based on metadata of still images or contents of still images. These methods were very useful for retrieving the original image without any distortion by using the pre-inputted information or the original information of the original still image (color, texture, etc.) of the original still image. However, if the still image is illegally deformed by copying or deliberate attack of the still image, or if the original characteristic (size, shape or quality of the image, etc.) of the still image changes depending on the type of Internet or playback device, It was impossible to apply the methods. In this case, a technique for identifying a deformed still image is newly required, and this technique is a still image identification technique.

As still image usage increases, still image identification technology will be newly developed, and still image identification technology will continue to increase as research continues now and in the future. However, the still image identification technology is not able to completely identify the modified still image in all situations, there are success probability and failure probability. Therefore, the quality is inevitably different depending on the still image identification software product. However, consumers cannot obtain objective assessments of this quality, and developers have not had the opportunity to identify and promote the superiority of their still image identification software products.

Accordingly, the present invention has been made to solve the above-mentioned problems of the prior art, and an object of the present invention is to evaluate the performance of still image identification techniques that are robust to various deformations of still images, and to obtain a certificate including the evaluation contents. To provide a still image identification software evaluation / certification system and method automatically issued on the system.

On the other hand, it is another object of the present invention to automatically process the standard of the evaluation system and certificate for the evaluation and certification of still image identification technologies on the system, and also to objectively and systematically compare and analyze the performance of the still image identification software. The present invention provides a still image identification software evaluation / authentication system and method that allows a user to select a technology suitable for his purpose based on the certificate contents of the still image identification technology.

Still image identification software evaluation / authentication system of the present invention for achieving the above object is a receiving module for receiving a still image identification software and the application for authentication associated with the still image identification software to be evaluated and certified, the received An identification test module for identifying the deformed still image data using still image identification software, an evaluation module for calculating and evaluating statistical information using the results identified by the identification test module, and statistical information and evaluation by the evaluation module Characterized in that it comprises a certificate generation module for generating a certificate using the result.

At this time, when the authentication application is received, characterized in that it further comprises an evaluation environment setting module for setting the evaluation environment suitable for the still image identification software.

The evaluation environment is characterized in that the evaluation criteria, evaluation items or evaluation scenarios are automatically set according to the application field suitable for the still image identification software.

The deformed still image data is obtained by deforming a still image stored in a still image database according to a deformation algorithm set according to the evaluation environment, and the deformation algorithm is characterized by being made of non-geometric deformation and geometric deformation.

The non-geometric deformations are characterized by brightness conversion, color conversion, compression ratio conversion, Gaussian noise value conversion, histogram smoothing conversion, and sharpness conversion. It is characterized by consisting of a distortion transformation.

The apparatus may further include a transmission module for transmitting a certificate generated by the certificate generation module to an external authority.

The certificate is characterized by consisting of performance evaluation information including information on the accuracy of identification, the complexity of the calculation, the size of the calculator, the accuracy of the identification is SR = K / (I + N) {SR is identification accuracy, K Is the number of transformed still images of the identified query image, I is the query image requiring identification, N is the number of transformed still images, and the complexity of the calculation is calculated by the amount of instructions and memory used. Characterized in that.

Meanwhile, the still image identification software evaluation / authentication method of the present invention comprises the first step of receiving a still image identification software and an application for certification related to the still image identification software, which are subject to evaluation and authentication, by the identification test module. Step 2 of identifying the deformed still image data by using the still image identification software, 3 step of evaluating and evaluating statistical information by the evaluation module using the result identified by the identification test module, and statistics by the evaluation module And four steps of generating a certificate using the information and the evaluation result.

In this case, the first step may further include setting an evaluation environment suitable for the still image identification software when an authentication application is received, wherein the evaluation environment is based on an application field suitable for the still image identification software. However, the evaluation items or evaluation scenarios are automatically set.

The second step may further include generating the deformed still image data by deforming the still image stored in the still image database according to the deforming algorithm set by the evaluation environment.

In addition, the step 4 further comprises the step of transmitting the generated certificate to an external authority.

Hereinafter, a still image identification software evaluation / authentication system and method of the present invention will be described in detail with reference to the accompanying drawings.

1 is a view showing the configuration of a still image identification software evaluation / authentication system according to the present invention.

Still image identification software evaluation / authentication system according to the present invention (hereinafter referred to as "evaluation / authentication system") is a receiving module 110, evaluation environment setting module 130, identification test module 150, evaluation module ( 170), the certificate generation module 180, and the still image DB 120 and the converted image DB 140.

The receiving module 110 receives a still image identification software and an authentication application associated with the still image identification software. Still image identification software refers to software developed for identifying modified still images. The certification application includes certification body information, certification applicant information, and information about still image identification products. Refer to [Table 1] for the details of each listed in the certification application.

[Table-1.Certification Application]

Certificate Authority Information Certification Body Name Certificate Authority Address Certificate Authority Serial Number Certificate Applicant Information Certification applicant name Certificate Applicant Address Certificate Applicant Identification Number Still Image Identification Product General Information Still image identification product identification number Patent (optional) Algorithm Description

The certification body information consists of information such as the certification body name, certification body address and certification body serial number. The certification applicant information includes information such as the certification applicant name, the certification applicant address, and the certification applicant identification number. The still image identification product general information includes information such as a still image identification product identification number and an algorithm.

When the evaluation environment setting module 130 receives the authentication application from the receiving module 110, the evaluation environment setting module 130 automatically sets an evaluation environment such as evaluation criteria, evaluation items, and evaluation scenarios according to an application field suitable for the received still image identification software. Information about the application may be provided by the certification body.

The use cases of still image identification software are very extensive, such as copyright protection, digital libraries, illegal copy and copy protection, data authentication, copy control and fingerprinting. In addition, evaluation criteria and parameters can also be set automatically, depending on the application.

Meanwhile, according to an application field, types of still images and converted images stored in the still image DB 120 or the converted image DB 140, and evaluation items thereof may be variously changed, and may also be changed during evaluation.

The evaluation environment setting module 130 may additionally set an additional tool (tool) and an evaluation environment by a support system for evaluating still image identification software, if necessary.

Additional tools and evaluation environments by the support system include classification of still image identification software application field, setting appropriate evaluation criteria for the application field, evaluation environment (signal processing, DB tool) and statistical processing methodology development and environment (spreadsheet, Using statistical processing).

In addition, the support system for evaluation of still image identification software enables the evaluation applicant or certification body to easily set up their evaluation environment.

The still image database for the still image identification software evaluation test is composed of a still image DB (120) and a converted image DB (140).

The still image DB 120 is composed of a color image, which is used to calculate a primary threshold for calculating the identification rate of the still image identification software product. The threshold value of the algorithm for identifying the original input image by testing still images by the still image identification software product is calculated by adding the sum of the respective threshold values of the finally identified still images to 50,000 * (50,000-1) / 2 ( It can be defined as the value divided by (for 50,000 color images).

When the color image of the still image DB 120 is 50,000, the converted image DB 140 is composed of 2,000 images selected by analyzing the energy values of the spatial and frequency domains in the image among 50,000 still images. These 2,000 images are transformed using a transform algorithm and used as transformed image data to evaluate the performance of still image identification software.

Still image transformation algorithms can be largely divided into non-geometric and geometric transformations as follows.

end. Non-geometrical modification

1.1. Brightness change

Changes the brightness of the input still image in the following proportions.

± 20%, ± 10%, ± 5%

1.2. Black / monochrome conversion

The input still image is converted by the following formula.

I = 0.299R + 0.587G + 0.114B

Here, R, G, and B represent red, green, and blue components of colors, respectively.

1.3. JPEG compression

As JPEG compression, compression is performed for QF 65, 80, and 95. Here, the Joint Photographic Experts Group (JPEG) is a lossy compression method standard created for still images. This standard was established by ISO and ITU-T. In addition, QF (Quality Factor) is a method of measuring the quality of still images.

1.4. Color reduction

The number of bits representing the pixel color value of the input still image is 16 to 24 bits (R = 8 bits, G = 8 bits, B = 8 bits). Bits, decrement by 8 bits

1.5. Gaussian noise

Change the standard deviation of the input still image by 8.0, 4.5, or 2.5.

1.6. Histogram equalization

The histogram value of the input still image should have the same value for all components.

1.7. Blurring

Decreases the sharpness of the input still image using the following mask.

5 * 5, 3 * 3

I. Geometrical modification

2.1. Rotation

Rotates to the next angle of the input still image.

45˚, 25˚, 10˚

2.2. Scaling

The horizontal and vertical size of the input still image is reduced by the following ratio.

50%, 70%, 90%

2.3. Equilibrium translation

Move the position of the input still image by the next ratio.

40%, 20%, 10%

2.4. Flip

Switch the left and right of the input still image.

2.5. Aspect ratio change

Change the horizontal / vertical ratio of the input still image as follows, and apply 4.2.6 cropping transformation.

4: 3-> 16: 9 + crop

2.6. Crop

The aspect ratio and arbitrary size of the input still image are truncated as follows.

50%, 70%, 90%

2.7. Skew

Change the value of α by +/- 10 °, +/- 6 ° or +/- 2 ° using the following formula for the input still image.

는 뒤틀린 정지영상의 좌표를 나타낸다.Where α is the distortion angle, ( u, v ) is the coordinates of the input still image,

Represents the coordinates of the twisted still image.

2.8. Perspective

Change the θ value by +/- 10 °, +/- 6 ° or +/- 2 ° using the following formula.

는 입력 정지영상의 좌표를 나타낸다.here

Denotes the coordinates of the input still image.

는 항상 양의 값을 가진다. 또, 수평축과 수직축은 각각 X축, Y축으로 정의한다. 그리고

는 X축에 대한 회전각을,

과

는 입력 정지영상의 폭과 높이를 나타낸다. here,

Always has a positive value. In addition, the horizontal axis and the vertical axis are defined by the X axis and the Y axis, respectively. And

Is the rotation angle about the X axis,

and

Indicates the width and height of the input still image.

The identification test module 150 identifies the converted image according to the evaluation environment set by the evaluation environment setting module 130 using the still image identification software product. The identification work is tested according to a test still image, a still image transformation function, an evaluation criterion and an item, an evaluation scenario, etc., which are determined in an evaluation environment set by the evaluation environment setting module 130.

The evaluation module 170 performs statistical processing according to the evaluation criteria and the statistical processing method and the statistical environment set by the evaluation environment setting module 130 using the result of the identification operation on the converted image performed by the identification test module 150. Perform

The evaluation environment of the still image identification product performance portion set by the evaluation environment setting module 130 may vary depending on the application field of the still image identification software. In other words, the use cases of still image identification software are very broad, such as copyright protection, digital libraries, illegal copy and copy protection, data authentication, copy control and fingerprinting, and each evaluation criteria and items. parameters can also be automatically set to various applications.

In addition, the selection criteria and evaluation items of the still image modified image may be very broad and complicated depending on the application field, and may be changed during evaluation.

In particular, the following additional tools and environments are required for evaluation of still image identification software.

1) Classification of Application Fields of Still Image Identification Software

2) Establishment of Evaluation Criteria for Application

3) evaluation environment (eg signal processing, D / B tools)

4) Development of statistical methodology and environment (eg, using spreadsheets, statistical package, etc.)

Using the tool, the evaluation module 170 calculates statistics on identification reliability and performance based on the identification work result. In addition, the evaluation module 170 calculates information such as identification accuracy, calculation complexity, and size of the calculator of the still image identification software together with the information on the reliability.

The certificate generation module 180 generates a certificate including the statistical processing result by the evaluation module 170. Therefore, the certificate generation module 180 tests the result of the received still image identification software according to the test still image, the still image transformation function, evaluation criteria and items, evaluation scenarios, and the like set in the evaluation environment. Write according to

Table-2. Certificate Form]

Performance evaluation information included in the certificate consists of information on identification accuracy, computational complexity, and calculator size.

Identification accuracy is represented by SR = K / (I + N). Where SR is the Success Ratio, I is the query image requiring identification, N is the number of modified still images, and K is the number of modified still images of the identified query image (I).

Computational complexity indicates the computational complexity for query image identification of still image identification products subject to authentication, in terms of instructions and amount of memory used.

The calculator size indicates the size of the data representing the calculator in bits.

2 is a diagram showing an embodiment of a still image identification software evaluation / authentication system according to the present invention.

The evaluation / authentication system is provided to the evaluation / authentication system 200 of the certification authority through the Internet, and the still image identification software evaluation / certification viewer by the certification application 211 and the interface and communication protocol specified by the certification authority. Send and receive.

The certification body conducts an identification test on the application form and the still image identification software submitted by the evaluation system (Image Identifier Evaluation Engine, 200) according to the set environment, and then prints the certificate 270 including the evaluation information. Issue to the owner or applicant. The certification authority also establishes a certificate database and manages the issued certificates.

The evaluation / certification system 200 determines the evaluation criteria and evaluation criteria (Evaluation Criteria & Parameters, 262) and evaluation scenarios (263) according to the application cases (use cases 261) suitable for the received still image identification software product. Set the evaluation environment.

Once the evaluation environment is set, the still image identification module (251) of the evaluation / authentication system 200 is a still image stored in the still image DB (Test Image DB, 221) using the applied still image identification software. And (Modification Tool DB, 222) Identification of the converted image is performed.

The converted image is image data transformed according to a deformation algorithm 256 set according to an evaluation environment of a still image stored in a still image DB (Test Image DB) 221.

The evaluation module 252 calculates statistics on identification reliability and performance based on the identified results. In addition, the evaluation module 252 calculates information such as identification accuracy, calculation complexity, and size of the calculator of the still image identification software together with the information on the reliability.

The evaluation / certification system 200 creates a certificate in accordance with a predetermined certification application form 254 based on the results calculated by the evaluation module Image Identifier Evaluation Tool 252, and the like. To send.

3 is a flowchart illustrating a still image identification software evaluation / authentication method according to the present invention.

The still image identification software evaluation / authentication method is performed by the following steps. First, an application for certification associated with still image identification software and still image identification software to be evaluated and authenticated is received from an external certification applicant organization (certification applicant) (S301).

When the authentication application is received, the evaluation environment (evaluation DB, attack function, evaluation criteria, evaluation items, evaluation scenarios, etc.) according to the application field of the received still image identification software is automatically set (S302).

When the evaluation environment is set, the modified still image data is generated by transforming the test still image data according to various deformation algorithms (S303).

When the evaluation environment is set and the modified still image data is generated, the deformed still image is identified using the received still image identification software (S304).

Statistical information on the reliability of the still image identification software is calculated using the identified result (S305).

Finally, a certificate containing the calculated reliability statistical information is generated and transmitted to the external certification applicant (certification applicant) (S306).

The still image identification software evaluation / authentication method may be implemented as a computer-readable recording medium (CD-ROM, RAM, floppy disk, hard disk, magneto-optical disk, etc.) that records a program for execution on a computer. .

Although the present invention has been described in more detail with reference to some embodiments, the present invention is not necessarily limited to these embodiments, and various modifications can be made without departing from the spirit of the present invention.

As described above, the still image identification software evaluation / authentication system and method according to the present invention has the effect of objectively and systematically comparing and analyzing the performance of existing or future still image identification software products.

In addition, the still image identification software evaluation / authentication system and method according to the present invention provides a direction for developing excellent still image identification technology to technology developers, the consumer has the effect of selecting the desired still image identification technology have.

In addition, the present invention has the effect that can provide objectively verified mutual trust through the systematic evaluation and certification for still image identification software products.

Claims (17)

delete delete delete delete delete A still image DB for storing still image data for use in calculating a threshold for calculating an identification ratio of a still image identification software product; A converted image DB for storing the still image data converted using the non-geometric transformation including brightness conversion, color conversion, compression ratio conversion, Gaussian noise value conversion, histogram smoothing conversion, and sharpness conversion for the selected still image; A receiving module for receiving a still image identification software to be evaluated and authenticated and an authentication application relating to the still image identification software; An evaluation environment setting module configured to set an evaluation environment including signal processing and a DB tool in response to the still image identification software when the certification application is received; An identification test module for identifying modified still image data using the received still image identification software; An evaluation module for calculating statistical information using the results identified by the identification test module to evaluate the accuracy of identification defined as SR = K / (I + N); Still image identification software evaluation / authentication system comprising a certificate generation module for generating a certificate using the statistical information and the evaluation result by the evaluation module. (Where SR is identification accuracy, K is the number of modified still images of the identified query image, I is the query image requiring identification, and N is the number of modified still images.) A still image DB for storing still image data for use in calculating a threshold for calculating an identification ratio of a still image identification software product; A converted image DB for storing the still image data converted using the geometric transformation including the rotation transformation, the size transformation, the equilibrium movement, the left-right transformation, the cropping, and the distortion transformation, for the selected still image; A receiving module for receiving a still image identification software to be evaluated and authenticated and an authentication application relating to the still image identification software; An evaluation environment setting module configured to set an evaluation environment including signal processing and a DB tool in response to the still image identification software when the certification application is received; An identification test module for identifying modified still image data using the received still image identification software; An evaluation module for calculating statistical information using the results identified by the identification test module to evaluate the accuracy of identification defined as SR = K / (I + N); Still image identification software evaluation / authentication system comprising a certificate generation module for generating a certificate using the statistical information and the evaluation result by the evaluation module. (Where SR is identification accuracy, K is the number of modified still images of the identified query image, I is the query image requiring identification, and N is the number of modified still images.) delete delete delete delete delete delete delete delete delete delete

Images