KR101038520B1 - Color laser printer identification system and method using discrete wavelet transform and statistical analysis - Google Patents

Abstract

The present invention relates to a color laser printer discrimination system and method using discrete wavelet transform and statistical analysis, comprising: scanning means for receiving an output printed by a color printer and digitizing the output into color image information; The RGB pixel values included in the color image information are first filtered through a high resolution filter and a low resolution filter to be classified into a first high frequency filtered image (Hz) and a first low frequency filtered image (Lz), and the first high frequency filtered image ( Discrete wavelet converting means for performing second-order filtering through a high-resolution filter and a low-resolution filter to classify Hz into a second high frequency filtered image (Hz-Hz) and a second low frequency filtered image (Lz-Lz); Classifying the R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz), and the standard deviation of each of the classified R region image, G region image, and B region image, Feature point extraction means for extracting kurtosis, skewness, covariance and correlation coefficient; Standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to RGB pixel values are converted to vector coordinate values and output, and the groups are classified according to the similarity of the distribution of the output vector coordinate values. Vector conversion means for deriving vectors belonging to a maximum-margin hyperplane group among the vectors of kurtosis, kurtosis, skewness, covariance and correlation coefficient; And a printer corresponding to the vectors of standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group according to the training of the Support Vector Machine (SVM). A printer discrimination means for generating a discrimination model for deriving the received RGB pixel value and the printer information and toner information corresponding to the vector of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the applied RGB pixel value according to the discrimination model; .

Color, Laser, Printer, Discrimination, Discrete Wavelets, Support Vector Machines

Description

COLOR LASER PRINTER IDENTIFICATION SYSTEM AND METHOD BY ANALYZING STATISTICAL FEATURES ON DISCRETE WAVELET TRANSFORM}

The present invention relates to a color laser printer discrimination system and method using discrete wavelet transform and statistical analysis, and more particularly, to convert the printed output into digitized color image information, and to convert the result into discrete wavelet to vectorize the extracted feature points. The present invention relates to a technique for determining whether or not color image information has been output by any printer and toner.

The change to the information society due to the development of information and communication technology has caused the popularization of digital devices. Recently released digital devices perform better than the past digital devices, despite their low price. This makes it easier for general users to access digital devices, and as a result, the reliability of data generated by digital devices is becoming increasingly important.

Digital image forensic technology is a technology related to the reliability of data generated by digital devices, and it is specialized in image-related digital devices represented by digital cameras, scanners, and high resolution color printers, and professional image editing such as Photoshop and Illustrator. Due to the popularization of tools, easy forgery of images using them is becoming more and more. In contrast, the development of technology that can detect such forgery is still in its infancy.

As described above, forensic technology is divided into image acquisition equipment discrimination and image forgery detection technology. First, image acquisition equipment discrimination technology is a technology for determining what equipment is used to acquire an image from a given image. It aims to determine the exact model name level as well as the type of equipment used to acquire the image using only the given image.

In addition, image forgery detection technology is a technique for determining whether a given image has undergone a forgery modulation step. Forgery of digital images leaves no visible clues, but changes the statistical properties that make up the image. It aims to find the forged and altered parts by finding such changed statistical characteristics.

However, the printer discrimination techniques studied so far are mainly for black and white laser printers, and these techniques are applicable only to black and white printer documents that exclude color laser printers. Therefore, it is possible to know in which printer the printed color documents are printed. This means that it is not possible to distinguish between forgery and alteration documents.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to convert a printed output into digitized color image information, convert it into discrete wavelets, and vectorize the extracted feature points to obtain color image information. Its purpose is to determine whether it has been output by a printer and toner.

Color laser printer determination system by the discrete wavelet transform and statistical analysis of the present invention for achieving the above technical problem, the scanning means for receiving the output printed by the color printer to digitize the output into color image information; The RGB pixel values included in the color image information are first filtered through a high resolution filter and a low resolution filter to be classified into a first high frequency filtered image (Hz) and a first low frequency filtered image (Lz), and the first high frequency filtered image ( Discrete wavelet converting means for performing second-order filtering through a high-resolution filter and a low-resolution filter to classify Hz into a second high frequency filtered image (Hz-Hz) and a second low frequency filtered image (Lz-Lz); Classifying the R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz), and the standard deviation of each of the classified R region image, G region image, and B region image, Feature point extraction means for extracting kurtosis, skewness, covariance and correlation coefficient; Standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to RGB pixel values are converted to vector coordinate values and output, and the groups are classified according to the similarity of the distribution of the output vector coordinate values. Vector conversion means for deriving vectors belonging to a maximum-margin hyperplane group among the vectors of kurtosis, kurtosis, skewness, covariance and correlation coefficient; And to train the corresponding support vector machine (SVM) to determine the corresponding printer with the vectors of standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group. Printer discrimination means for generating a discrimination model and deriving printer information and toner information corresponding to a vector of standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the applied RGB pixel value according to the discrimination model; .

In addition, the color laser printer determination system by the discrete wavelet transform and statistical analysis of the present invention, the scanning means for receiving the output printed by the color printer to digitize the output into color image information; The CMYK pixel values included in the color image information are first filtered through a high resolution filter and a low resolution filter to classify the third high frequency filtered image (Hz) and the third low frequency filtered image (Lz), and classify the high resolution image (Hz). Discrete wavelet converting means for performing second-order filtering through a high resolution filter and a low resolution filter to classify the fourth high frequency filtered image (Hz-Hz) and the fourth low frequency filtered image (Lz-Lz); Classify the C region image, the M region image, the Y region image, and the K region image corresponding to the CMYK pixel value of the fourth high frequency filtered image (Hz-Hz), and classify the classified C region image, M region image, Y region image, and K region. Feature point extraction means for extracting standard deviation, kurtosis, skewness, covariance and correlation coefficient of each region image; Standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to CMYK pixel values are converted to vector coordinate values, and the groups are classified according to the similarity of the distribution of the output vector coordinate values. Vector conversion means for deriving vectors belonging to a maximum-margin hyperplane group among the vectors of kurtosis, kurtosis, skewness, covariance and correlation coefficient; And a corresponding vector of standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to CMYK pixel values of the maximum margin hyperplane group according to training of the Support Vector Machine (SVM). Printer discrimination means for generating a discrimination model and deriving printer information and toner information corresponding to a vector of standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to the CMYK pixel value again, according to the discrimination model; .

Hereinafter, the method for discriminating color laser printers by the discrete wavelet transform and statistical analysis of the present invention based on the above-described system may include: (a) receiving the output printed by the color printer and converting the color output into color image information through digitization; ) step; The discrete wavelet transform means classifies the RGB pixel values included in the color image information into a first high frequency filtered image (Hz) and a first low frequency filtered image (Lz) by performing first-order filtering through a high resolution filter and a low resolution filter (b). ) step; The discrete wavelet transform unit performs second-order filtering on the first high frequency filtered image (Hz) and the first low frequency filtered image (Lz) through a high resolution filter and a low resolution filter, thereby performing a second high frequency filtered image (Hz-Hz) and a second. (C) classifying the low frequency filtered image (Lz-Lz); The discrete wavelet transform means classifies the CMYK pixel value included in the color image information into a third high frequency filtered image (Hz) and a third low frequency filtered image (Lz) by performing first-order filtering through a high resolution filter and a low resolution filter (d). ) step; Discrete wavelet transform means performs second-order filtering on the third high frequency filtered image (Hz) and the third low frequency filtered image (Lz) through a high resolution filter and a low resolution filter to perform a fourth high frequency filtered image (Hz-Hz) and a fourth low frequency. (E) classifying the filtered image into Lz-Lz; The feature point extracting means classifies the R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz), and the standard of each of the R region image, the G region image, and the B region image. (F) extracting deviation, kurtosis, skewness, covariance and correlation coefficient; The feature point extraction unit classifies the C region image, the M region image, the Y region image, and the K region image corresponding to the CMYK pixel value of the fourth high frequency filtered image (Hz-Hz), and the C region image, the M region image, and the Y region image. (G) extracting the standard deviation, kurtosis, skewness, covariance and correlation coefficient of each of the image and the K region image; The vector converting means converts the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to RGB pixel values, and the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to CMYK pixel values into a vector (H) outputting a coordinate value; The vector conversion means classifies the output vector coordinate values according to the similarity of the vector coordinate value distribution into two groups, and corresponds to the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values, and the CMYK pixel values. (I) deriving vectors belonging to the maximum margin hyperplane group among the vectors of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient, respectively; (j) Through the training of the support vector machine, the printer discrimination means corresponds to the vectors of standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group, and the CMYK pixel values. Generating a discrimination model for discriminating a printer corresponding to vectors of standard deviation, kurtosis, skewness, covariance, and correlation coefficient; And vectors of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the RGB pixel value again supplied by the printer discrimination means, and the standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the CMYK pixel value. (K) deriving printer and toner information corresponding to the vector according to the discrimination model; It includes.

According to the present invention as described above, by converting the printed output into digitized color image information, and converting it into discrete wavelets to vectorize the extracted feature points, as to determine whether the color image information is output by what printer and toner There is an effect of discriminating forgery or alteration of printed output.

Specific features and advantages of the present invention will become more apparent from the following detailed description based on the accompanying drawings. Prior to this, terms and words used in the present specification and claims are to be interpreted in accordance with the technical idea of the present invention based on the principle that the inventor can properly define the concept of the term in order to explain his invention in the best way. It should be interpreted in terms of meaning and concept. In addition, when it is determined that the detailed description of the known function and its configuration related to the present invention may unnecessarily obscure the subject matter of the present invention, it should be noted that the detailed description is omitted.

1 is a block diagram showing a color laser printer discrimination system S by discrete wavelet transform and statistical analysis according to the present invention, as shown, scanning means 100, discrete wavelet transform means 200, and feature point extraction means. 300, the vector converting means 400, and the printer discriminating means 500.

As shown in FIG. 1, the scanning means 100 receives an output 10 printed by a color printer, and digitizes the pattern printed on the output into color image information.

Meanwhile, as illustrated in FIG. 2, the discrete wavelet converting means 200 may include values of RGB (Read, Green, Blue) pixels included in the color image information applied from the scanning means 100 (hereinafter, referred to as 'RGB pixel values'). 'Are first filtered through the high resolution filter 210 and the low resolution filter 220 to be classified into a first high frequency filtered image Hz and a first low frequency filtered image Lz, and a first high frequency filtered image Hz. ) Is again filtered through the high resolution filter 210 and the low resolution filter 220 to classify the second high frequency filtered image (Hz-Hz) and the second low frequency filtered image (Lz-Lz).

In addition, the discrete wavelet converting means 200 may determine the patterns included in the color image information received from the scanning means 100 in CMYK (Cyan, Magenta, Yellow, Key (Black)) pixel values (hereinafter, 'CMYK pixel value'). ) Are first filtered through the high resolution filter 210 and the low resolution filter 220 to be classified into a third high frequency filtered image Hz and a third low frequency filtered image Lz, and the classified third high frequency filtered image Hz is again filtered through the high resolution filter 210 and the low resolution filter 220 to classify the fourth high frequency filtered image (Hz-Hz) and the fourth low frequency filtered image (Lz-Lz).

는 아래와 같이 이산 웨이블릿의 기저함수

와 웨이블릿 계수

가 결합된 형태로 표현되며, 이는 아래의 [수학식1] 및 [수학식2]와 같다.In addition, if the equation according to the first order filtering and the second order filtering of the discrete wavelet transform means 200, an arbitrary signal

The base function of the discrete wavelet is

And wavelet coefficients

Are expressed in a combined form, which are represented by Equations 1 and 2 below.

[Equation 1]

&Quot; (2) "

와 웨이블릿 함수

가 존재하며, 이는 아래의 [수학식3]과 같이 정의된다.Also, for multi-resolution analysis using discrete wavelets, two basic functions, namely scale functions, are used.

And wavelet functions

Is present, and is defined as in Equation 3 below.

&Quot; (3) "

Meanwhile, the feature point extracting means 300 classifies the R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz) applied from the discrete wavelet transform means 200. Extracts the covariance and correlation coefficient between the R region image, G region image, G region image and B region image, and R region image and B region image, and estimates the average The standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each of the R region, G region, and B region images are extracted.

In addition, the feature point extracting means 300 is a C region image, M region image, Y region image and K region corresponding to the CMYK pixel value of the fourth high frequency filtered image (Hz-Hz) applied from the discrete wavelet transform means 200. Classify the image, C area image and M area image, M area image and Y area image, Y area image and K area image, C area image and Y area image, C area image and K area image and M area image and K area Extract the covariance and correlation coefficient between each region of the region images, and standard deviation, kurtosis, skewness, covariance, and correlation of each of the C region image, M region image, Y region image, and K region image except the average value Extract the coefficients

The 'standard deviation' for the R region image, G region image, B region image, C region image, M region image, Y region image, and K region image as described above is derived through Equation 4 below.

&Quot; (4) "

In addition, the kurtosis of the R region image, the G region image, the B region image, the C region image, the M region image, the Y region image, and the K region image is derived through Equation 5 below.

[Equation 5]

In addition, the 'distortion' of the R region image, G region image, B region image, C region image, M region image, Y region image, and K region image is derived through Equation 6 below.

[Equation 6]

In addition, 'covariance' for the R region image, the G region image, the B region image, the C region image, the M region image, the Y region image, and the K region image is derived through Equation 7 below.

[Equation 7]

In addition, the 'correlation coefficient' for the R region image, the G region image, the B region image, the C region image, the M region image, the Y region image, and the K region image is derived through Equation 8 below.

[Equation 8]

The vector converting means 400 is a standard deviation, kurtosis, skewness, covariance and correlation coefficient corresponding to the RGB pixel value extracted by the feature point extracting means 300, and standard deviation, kurtosis, why corresponding to the CMYK pixel value. Each of the degree, covariance, and correlation coefficient is converted into a vector and output as coordinates, and the two groups are classified according to the similarity of the distribution of coordinate values (hereinafter, 'vector coordinate values') for the output vector. The maximum-margin hyperplane of each of the vectors of the standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to, and the standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to CMYK pixel values. Deriving the vectors belonging to the group.

At this time, the derivation of the vectors belonging to the maximum margin hyperplane group, as shown in Figure 3, two groups classified according to the similarity distribution ('vector of the first group (●)' and 'vector of the second group ( ○) ') set the maximum margin hyperplane along the coordinates of the farthest orthogonal distance from the coordinate values located closest to the group among the vector coordinate values, respectively, and based on the set maximum margin hyperplane, The standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to the RGB pixel values and the standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to the CMYK pixel values are derived, respectively.

3 denotes a minimum margin hyperplane, and H2 denotes a maximum margin hyperplane.

As shown in FIG. 4, the printer discrimination means 500 is a RGB pixel value of the maximum margin hyperplane group from the vector conversion means 400 through a training process of a support vector machine (SVM) 510. Discrimination to determine the corresponding vector of the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to and the vector of the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to CMYK pixel values A vector of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the RGB pixel value that is again applied according to the discriminant model generated by the support vector machine 510, that is, the model of the trained support vector machine. And printer and toner information 20 corresponding to vectors of standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to CMYK pixel values.

5A and 5B are diagrams showing that the printer discriminating means 500 according to the present invention derives a printer that matches the color image information digitized by the scanning means 100 and the original color image information. 6A and 6B are diagrams showing the result of the printer discriminating means 500 according to the present invention deriving the toner matching the color image information digitized by the scanning means 100 and the original color image information.

As described above, the printed output is converted into digitized color image information, converted into discrete wavelets, and extracted feature points are converted into vectors, and the digitized color image information is generated according to the discrimination model generated by the support vector machine 510. Can be used to determine which printers and toners have been output, and to derive the forgery and alteration against the original color image information.

A method of transmitting caller information using the above-described system S will be described with reference to FIG. 7 as follows.

7 is a flowchart illustrating a method for transmitting caller information according to the present invention, the scanning means 100 receives an output 10 printed by a color printer and converts it into color image information through digitization (S100). ).

Subsequently, the discrete wavelet converting unit 200 performs first-order filtering on the RGB pixel values included in the color image information received from the scanning unit 100 through the high resolution filter 210 and the low resolution filter 220 to perform a first high frequency. The image is classified into a filtered image Hz and a first low frequency filtered image Lz (S200).

Subsequently, the discrete wavelet transform unit 200 performs second order filtering on the first filtered high frequency filtered image Hz and the second low frequency filtered image Lz through the high resolution filter 210 and the low resolution filter 220. In operation S300, the method may be classified into a second high frequency filtered image (Hz-Hz) and a second low frequency filtered image (Lz-Lz).

Behind the scenes, the discrete wavelet transform means 200 performs a first order filtering on the CMYK pixel values included in the color image information received from the scanning means 100 through the high resolution filter 210 and the low resolution filter 220 to perform a third high frequency. The filter is classified into a filtered image Hz and a third low frequency filtered image Lz (S400).

Subsequently, the discrete wavelet transform unit 200 performs second-order filtering on the first-high-filtered third high-frequency filtering image Hz and the third low-frequency filtered image Lz through the high resolution filter 210 and the low resolution filter 220. In operation S500, the method is classified into a fourth high frequency filtered image (Hz-Hz) and a fourth low frequency filtered image (Lz-Lz).

The trailing point, the feature point extractor 300 classifies the R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz) applied from the discrete wavelet transform means 200. The standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each of the R region image, the G region image, and the B region image are extracted (S600).

Subsequently, the feature point extracting means 300 corresponds to the CMYK pixel value of the fourth high frequency filtered image (Hz-Hz) applied from the discrete wavelet converting means 200. The C region image, the M region image, the Y region image, and the K region The image is classified, and standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each of the C region image, the M region image, the Y region image, and the K region image are extracted (S700).

Subsequently, the vector converting means 400 includes the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values extracted by the feature point extracting means 300, and the standard deviation, kurtosis, corresponding to the CMYK pixel values. The skewness, covariance, and correlation coefficients are converted into vectors to output respective vector coordinate values (S800).

Behind the scenes, the vector converting means 400 classifies the two groups according to the similarity of the distribution of the vector coordinate values, and the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values, and CMYK pixel values Among the vectors of the corresponding standard deviation, kurtosis, skewness, covariance, and correlation coefficient, vectors belonging to the maximum margin hyperplane group are derived (S900).

Subsequently, the printer discrimination means 500 performs a training process of the support vector machine 510, and the standard deviation, kurtosis, skewness, covariance, and the like corresponding to the RGB pixel values of the maximum margin hyperplane group from the vector conversion means 400. A discrimination model is generated to discriminate the vectors corresponding to the vectors of the correlation coefficient and the vectors of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the CMYK pixel values (S1000).

Then, the printer discrimination means 500 is a vector of the standard deviation, kurtosis, skewness, covariance and correlation coefficient corresponding to the RGB pixel value which the support vector machine 510 is applied again, and the standard deviation corresponding to the CMYK pixel value. The printer and toner information 20 corresponding to the vector of kurtosis, kurtosis, skewness, covariance, and correlation coefficient are derived according to the generated discrimination model, that is, the model of the supported support vector machine (S1100).

As described above and described with reference to a preferred embodiment for illustrating the technical idea of the present invention, the present invention is not limited to the configuration and operation as shown and described as described above, it is a deviation from the scope of the technical idea It will be understood by those skilled in the art that many modifications and variations can be made to the invention without departing from the scope of the invention. Accordingly, all such suitable changes and modifications and equivalents should be considered to be within the scope of the present invention.

1 is a block diagram showing a color laser printer identification system (S) by discrete wavelet transform and statistical analysis according to the present invention.

2 is an exemplary view showing filtering RGB and CMYK images by the discrete wavelet transform means 200 of the color laser printer discrimination system S by the discrete wavelet transform and statistical analysis according to the present invention.

3 is an exemplary diagram showing vectors belonging to a maximum margin hyperplane group according to the color laser printer discrimination system S by the discrete wavelet transform and statistical analysis of the present invention;

4 is a view showing original color image information and color image information for each printer by the printer discrimination means 500 of the color laser printer discrimination system S by discrete wavelet transform and statistical analysis according to the present invention.

5A and 5B show the color image information digitalized by the scanning means 100 and the original color of the printer discrimination means 500 of the color laser printer discrimination system S by discrete wavelet transform and statistical analysis according to the present invention. A diagram showing the derivation of printers that match each other among image informations.

6A and 6B show the color image information digitalized by the scanning means 100 and the original color of the printer discriminating means 500 of the color laser printer discrimination system S by discrete wavelet transform and statistical analysis according to the present invention. A diagram showing the results of deriving toners consistent with each other among image informations.

7 is a flowchart illustrating a method for determining a color laser printer by discrete wavelet transform and statistical analysis according to the present invention.

** Description of symbols for the main parts of the drawing **

S: Color Laser Printer Discrimination System by Discrete Wavelet Transform and Statistical Analysis

100: scanning means 200: discrete wavelet converting means

300: feature point extraction means 400: vector conversion means

500: printer discrimination means 510: support vector machine

10: printed output 20: printer and toner information

Claims (7)

In the color laser printer identification system by discrete wavelet transform and statistical analysis, Scanning means for receiving an output printed by a color printer and digitizing the output into color image information; The RGB pixel values included in the color image information are first filtered through a high resolution filter and a low resolution filter to be classified into a first high frequency filtered image Hz and a first low frequency filtered image Lz, and the first high frequency filtering. Discrete wavelet converting means for classifying an image (Hz) into a second high frequency filtered image (Hz-Hz) and a second low frequency filtered image (Lz-Lz) by performing second-order filtering through the high resolution filter and the low resolution filter; The R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz) are classified, and the standard deviation of each of the classified R region image, G region image, and B region image Feature point extraction means for extracting kurtosis, skewness, covariance, and correlation coefficient; The standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the RGB pixel values are converted into vector coordinate values, and outputted. The groups are classified according to the similarity of the output vector coordinate value distribution to correspond to the RGB pixel values. Vector conversion means for deriving vectors belonging to a maximum-margin hyperplane group among vectors of standard deviation, kurtosis, skewness, covariance, and correlation coefficient; And Under the training of a Support Vector Machine (SVM) to determine the corresponding printer with the vectors of the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group A printer which generates a discrimination model and derives printer information and toner information corresponding to a vector of standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group according to the discrimination model Discriminating means; Color laser printer identification system by the discrete wavelet transform and statistical analysis comprising a. The method of claim 1, The average value of each of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each of the R region, G region, and B region images is determined to be noise. Color laser printer identification system by discrete wavelet transform and statistical analysis characterized in that the filtering. The method of claim 1, Derivation of the vector belonging to the maximum margin hyperplane group, The RGB margin pixel is set based on the largest margin hyperplane along the coordinates of the furthest distance from the output vector coordinate values, the coordinates of which are located closest to the group. A color laser printer discrimination system using discrete wavelet transform and statistical analysis, which derives a standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to a value. In the color laser printer identification system by discrete wavelet transform and statistical analysis, Scanning means for receiving an output printed by a color printer and digitizing the output into color image information; The CMYK pixel values included in the color image information are first filtered through a high resolution filter and a low resolution filter to be classified into a third high frequency filtered image (Hz) and a third low frequency filtered image (Lz), and the high resolution image (Hz). Discrete wavelet converting means for performing second-order filtering through the high resolution filter and the low resolution filter to classify a fourth high frequency filtered image (Hz-Hz) and a fourth low frequency filtered image (Lz-Lz); Classifying the C region image, the M region image, the Y region image, and the K region image corresponding to the CMYK pixel value of the fourth high frequency filtered image (Hz-Hz), and the classified C region image, M region image, Y region image, Feature point extraction means for extracting standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each K-region image; The standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the CMYK pixel value are converted into vector coordinate values, and outputted. The groups are classified according to the similarity of the output vector coordinate value distribution to correspond to the CMYK pixel values. Vector conversion means for deriving vectors belonging to a maximum-margin hyperplane group among vectors of standard deviation, kurtosis, skewness, covariance, and correlation coefficient; And Under the training of a Support Vector Machine (SVM) to determine the printer corresponding to the vectors of the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the CMYK pixel values of the maximum margin hyperplane group Printer discrimination means for generating a discrimination model and deriving printer information and toner information corresponding to vectors of the standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to the CMYK pixel values again, according to the discrimination model; Color laser printer identification system by the discrete wavelet transform and statistical analysis comprising a. The method of claim 4, wherein The mean values of the standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each of the C region image, the M region image, the Y region image, and the K region image are noise Color laser printer discrimination system by discrete wavelet transform and statistical analysis characterized in that it is determined by the filtering. The method of claim 4, wherein Derivation of the vector belonging to the maximum margin hyperplane group, The CMYK pixel is set based on the maximum margin hyperplane along the coordinates of the furthest distance from the orthogonal distances with the coordinate values located closest to the group among the output vector coordinate values. A color laser printer discrimination system using discrete wavelet transform and statistical analysis, which derives a standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to a value. In the color laser printer discrimination method by discrete wavelet transform and statistical analysis, (a) scanning means receiving the output printed by the color printer and converting the color image information through digitization; (b) the discrete wavelet converting means performs first-order filtering on the RGB pixel values included in the color image information through a high-resolution filter and a low-resolution filter to convert the first high frequency filtered image (Hz) and the first low frequency filtered image (Lz). Classifying; (c) the discrete wavelet converting means performs second-order filtering on the first high frequency filtered image Hz and the first low frequency filtered image Lz through the high resolution filter and the low resolution filter to generate a second high frequency filtered image Hz-. Hz) and a second low frequency filtered image (Lz-Lz); (d) the discrete wavelet converting means performs first-order filtering on CMYK pixel values included in the color image information through a high resolution filter and a low resolution filter to generate a third high frequency filtered image (Hz) and a third low frequency filtered image (Lz). Classifying to; (e) the discrete wavelet converting means performs second-order filtering on the third high frequency filtered image Hz and the third low frequency filtered image Lz through the high resolution filter and the low resolution filter to generate a fourth high frequency filtered image Hz-. Hz) and a fourth low frequency filtered image (Lz-Lz); (f) the feature point extracting means classifies the R region image, the G region image, and the B region image corresponding to the RGB pixel values of the second high frequency filtered image (Hz-Hz), and the R region image, the G region image, and the B region image. Extracting the standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each region image; (g) the feature point extracting means classifies the C region image, the M region image, the Y region image, and the K region image corresponding to the CMYK pixel value of the fourth high frequency filtered image (Hz-Hz), wherein the C region image, Extracting the standard deviation, kurtosis, skewness, covariance, and correlation coefficient of each of the M region image, the Y region image, and the K region image; (h) the vector converting means converts the standard deviation, kurtosis, skewness, covariance and correlation coefficient corresponding to the RGB pixel value and the standard deviation, kurtosis, skewness, covariance and correlation coefficient corresponding to the CMYK pixel value into a vector Converting and outputting each vector coordinate value; (i) the vector converting means classifies two groups of outputted vector coordinate values according to the similarity of the distribution of the vector coordinate values, and includes the standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to the RGB pixel values. Deriving vectors belonging to a maximum margin hyperplane group among the vectors of standard deviation, kurtosis, skewness, covariance, and correlation coefficient corresponding to CMYK pixel values; (j) The printer discrimination means corresponds to the vectors of the standard deviation, kurtosis, skewness, covariance and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group and the CMYK pixel values through the training of the support vector machine. Generating a discrimination model for discriminating the printer corresponding to the vectors of the standard deviation, the kurtosis, the skewness, the covariance, and the correlation coefficient; And (k) the printer discriminating means determines the vectors of standard deviation, kurtosis, skewness, covariance, and correlation coefficients corresponding to the RGB pixel values of the maximum margin hyperplane group, and the standard deviation, kurtosis, why corresponding to CMYK pixel values. Deriving printer and toner information corresponding to a vector of covariance and correlation coefficient according to the discrimination model; Color laser printer discrimination method by discrete wavelet transform and statistical analysis comprising a.

Images