JP2003333354A - Image processing method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To facilitate the adjustment of image processing by each model of input output apparatuses. <P>SOLUTION: The image processing method and program stores a parameter for defining a color space wherein a particular image exists to a storage means in the image processing, processes input image data by using the parameter stored in the storage means, then compares processing data of the input image data with ideal processing data of the input image data and corrects the parameter so that the processing data of the input image data are in matching with the ideal processing data on the basis of the comparison result. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a specific image recognition process.

[0002]

2. Description of the Related Art In recent years, the functions and performances of color copying machines have improved, and the prevention of counterfeiting of banknotes has become a serious problem, and effective anti-counterfeiting methods are being studied. In one of the anti-counterfeiting methods, a specific pattern is put in advance in a pattern such as a bill. When a color image is read at the time of copying in a copying machine, the scanned image is analyzed. In the image recognition process, if a specific pattern is detected in the color image, it is determined that a bill or the like will be copied, and normal image formation is prohibited.

A binarization process is often used as an image process for recognizing a specific pattern in a color image.
For example, when the specific pattern exists in a predetermined color space, it is binarized by determining whether or not the color image data exists in that color space. For this processing, the parameters defining the color space are stored in a dictionary, and the dictionary is used to binarize the color image data.

[0004]

The characteristics of input / output devices such as a scanner for inputting image data and a printer for printing image data on paper differ from device to device. Therefore, it is necessary to adjust these devices, the output data thereof, and the image processing relating to the printed matter for accurate processing.
For example, it is desirable to adjust the above-mentioned parameters used for image recognition for each model. However, such adjustment work can be performed only by an expert, and the adjustment work requires a great deal of time. Therefore, in order to perform image processing accurately, automation of adjustment work is an important issue.

An object of the present invention is to facilitate adjustment of image processing for each model.

[0006]

An image processing program executed by a computer according to the present invention stores a parameter defining a color space in which a specific image exists in a storage means, and stores the parameter in the storage means. Processing the input image data using the parameters, the step of comparing the processed data of the input image data with the reference data for the input image data, and the processed data of the input image data based on the result of the comparison. Modifying the parameters so that the (binarized data) matches the reference data. The processing of the input image data includes, for example, a binarization processing for determining whether the input image data is within the defined color space. This binarization may include a comprehensive judgment from the processing result of the input image data in a plurality of pixels for the same image. The parameters defining the color space may include parameters related to the color of the specific image itself as well as parameters related to the background color of the specific image. Also, for example,
The input image data is read data of a sheet including a plurality of color patches, and the reference data is reference data obtained by reading the sheet.

In the above image processing program, preferably, the processing step, the comparison step and the correction step are repeated until the processed data of the input image data matches the reference data.

The above-mentioned image processing program preferably further comprises a step of inputting image data from an image input device for reading an image.

The above-mentioned image processing program preferably further comprises a step of storing the processed data of the input image data in the storage means.

In the image processing method according to the present invention, the parameter defining the color space in which the specific image exists is stored in the storage means, the input image data is processed using the parameter stored in the storage means, and the input image data is processed. The processed data of the data is compared with the reference data for the input image data, and based on the result of the comparison, the parameters are modified so that the processed data of the input image data matches the reference data.

An image processing apparatus according to the present invention is a storage unit for storing parameters defining a color space in which a specific image exists, and a processing unit for processing input image data using the parameters stored in the storage unit. And comparing means for comparing the processed data of the input image data with the reference data for the input image data, and the processed data of the input image data to match the reference data based on the result of the comparison by the comparing means. And correction means for correcting the above parameters.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 shows the configuration of an image processing apparatus (hereinafter referred to as a system) according to an embodiment of the present invention. This system is a CPU that controls the entire system
It is composed mainly of 10. The CPU 10 is connected via a data bus to a ROM 12 in which an image processing program and the like are stored and a RAM 14 in which data and the like are stored. Furthermore, a display 18 for performing various displays
Is connected to the CPU 10 via the display control circuit 16. Keyboard 22 for performing various inputs and instruction operations
And mouse 26 are keyboard 22 and mouse 2, respectively.
4 is connected to the CPU 10 via a keyboard control circuit 20 and a mouse control circuit 24 which transfer control the input from the CPU 4.
The flexible disk 30 and the hard disk 34 are connected to the CPU 10 via a control circuit 28 for controlling the flexible disk device and a control circuit 32 for controlling the hard disk device, respectively. A printer 38 for printing image data is connected to the CPU 10 via a printer control circuit 36. The scanner 42 for taking in the image data is a CPU via the scanner control circuit 40.
Connected to 10.

In this system, the image processing program including the dictionary automatic adjustment function according to the present invention is stored in the ROM 12, but this program may be read and executed by a device such as a personal computer. A flexible disk 30 is used as a storage medium for dictionary data.
However, this may be another information storage medium. The dictionary data is read into the RAM 14, for example.
Although the scanner 42 is used as the image data input device, other image input devices such as the CD-ROM 12 and a digital camera (not shown) may be used. Although the printer 38 is used as the output device, another output device such as a digital copier may be used. The dictionary automatic adjustment function may be incorporated in a copying machine or the like.

First, an example of the binarization process which is one of the image recognition processes will be described. Here, the RG shown in FIG.
It is assumed that the absolute color range in the B space is the color space in which the specific image exists. Binarization is performed depending on whether multi-valued image data is in this color space. Here, the pixel whose red, green, and blue gradation values (R, G, B) of the pixel of interest satisfy all of the following first and second conditions are called ON pixels (≠ 0), When the condition is not met, it is called off pixel (= 0).

The first condition is that the pixel is in the RGB space shown in FIG. That is, R of the pixel of interest
The gradation value (brightness) of GB satisfies the following formula. Here, i is a variable indicating a mark (pixel of interest). R _imin ≤ R ≤ R _imax G _imin ≤ G ≤ G _imax B _imin ≤ B ≤ B _imax diffRB _imin ≤ R-B ≤ diffRB _imax diffRG _imin ≤ R-G ≤ diffRG _imax diffGB _imin ≤ G-B ≤ diffGB _imax

The second condition is the difference contR, cont with the background color.
It is about G and contB. This is because when a specific image exists in a specific background, in addition to the first condition for detecting the existence of the specific image, the difference between the specific image and the background can be considered for detection. This is because it is suitable for recognition. The background color may be measured, for example, by using the first patch as the background color or using a patch with the background color. contR, contG, contB are the gradation values of the pixel of interest
It is the difference between (R, G, B) and the maximum value (R _bmax , G _bmax , B _bmax ) of the gradation value (R, G, B) of the surrounding pixels (background color), and is _calculated as follows. To be contR = α × R _bmax −R contG = α × G _bmax −G contB = α × B _bmax −B The parameter α is, for example, 1, but is not limited thereto. The second condition consists of the following equation. _{contR imin ≦ contR ≦ contR imax contG} imin ≦ contG ≦ contG imax contB imin ≦ contB ≦ contB imax contdiffRB imin ≦ contR-contB ≦ contdiffRB imax contdiffRG imin ≦ contR-contG ≦ contdiffRG imax contdiffGB imin ≦ contG-contB ≦ contdiffGB imax

Table 1 shows a specific structure of a dictionary that stores parameters used for the first and second conditions.

[Table 1]

Needless to say, when the background color is not taken into consideration, only the first condition is necessary. The binarization condition is not limited to the example of FIG. 2 described above. In other cases, a corresponding relational expression is defined and its parameters are stored in the dictionary. The dictionary is input from an ordinary recording medium such as a flexible disk.

Next, the dictionary adjustment for the input system device will be described with reference to FIG. The input system device here is the scanner 42. First, the color patch chart is read by the scanner 42 (S10), and its electronic data (multi-valued image data) is input from the scanner 42 (S12). On the other hand, the image recognition dictionary in the storage unit stores parameters defining the color space in which the specific image exists (S14). The electronic data is processed using this dictionary (S16), and the binary image of the color patch chart is stored (S18). The processing of the electronic data includes processing of determining whether or not the electronic data of each pixel exists in the color space in which the specific image exists. here,
Binary image data is obtained depending on whether or not it exists in the color space. Next, the obtained processed data (binary image data) is compared with the binary information of the ideal color patch chart (S22). This "binary information of the ideal color patch chart" is reference data manually created using the above-mentioned color patch chart using a reference scanner (not shown), and is stored in advance in the storage means. (S20). Therefore, the binary image data is compared with this ideal processed data (reference data) to determine whether the data match (S24). If they do not match, the parameters defining the color space defined in the dictionary are modified so as to match the ideal color patch chart binary information (S26), and the result is reflected in the dictionary. A specific example of the modification will be described later. Then, the above-described electronic data processing and comparison (S16 to S24) are repeated until the processed data of the input image data matches the ideal processed data.
As described above, since the dictionary is automatically adjusted using the reference data for the color patch chart, the effort for creating the dictionary is reduced.

After the dictionary is adjusted in this way, when the actual read image data is input and the specific image is recognized,
Since the adjusted dictionary is used, the pixels in the color space corresponding to the specific image can be accurately detected. Then, based on that, it is recognized whether or not the specific image exists. A description of this recognition processing is omitted.

In the color patch chart shown in FIG. 4, each patch has a size of 3 × 3 pixels. With respect to the electronic data read from the color patch chart, the data of 9 pixels in each patch is binarized using the parameters of the dictionary. Then, the on (off) pixels in each patch are counted. Pat1_On = 8 Pat2_On = 5 Pat3_On = 6 ・ ・ ・ where Pati represents the i-th patch, and Pati
_On represents the number of ON pixels of the i-th patch. If the number of ON pixels is 6 or more for each patch, the patch is 1
(On), 0 (off) if less. Target_pati indicates whether the number of ON pixels for the i-th patch is 6 or more or less. In the above example: Target_pat1 = 1 Target_pat2 = 0 Target_pat3 = 1 ...

On the other hand, the ideal binary information obtained in advance is used to obtain the on / off information Ideal_pati for the color patch chart. Here, Ideal_pati indicates whether the number of ON pixels for the i-th patch is 6 or more or less. If the number of ON pixels in the patch is 6 or more, it is set to 1 (ON), and if it is less than 0, it is set to 0 (OFF). Ideal_pat1 = 1 Ideal_pat2 = 0 Ideal_pat3 = 1 ・ ・ ・

Also, the average color information (R _ave , G _ave ,
Calculate B _ave ). For example, for patch N: PatN_R _ave = 250 PatN_G _ave = 182 PatN_B _ave = 75

Next, for each patch, the processing result Target_pati using the dictionary is compared with the processing result Ideal_pati using the binarized information of the ideal color patch chart, and the parameters of the dictionary are corrected according to the result. Here, the update order of the parameters is determined in the following order. Here, diffRG = R-G, diffRB = RB, and diffGB = GB. Further, the amounts contGB and diffcontGB to which cont is added are amounts related to the difference from the peripheral pixels.

(A) Nth patch Idea1_patN =
When 1 and Target_patN = 0, that is, when an on pixel is mistakenly recognized as an off pixel, the color space parameter is adjusted to the low gradation level side as follows.

(B) Ideal_patN = for the Nth patch
When 0 and Target_patN = 1, that is, when an off pixel is erroneously recognized as an on pixel, the color space parameter is adjusted to the high gradation side as follows.

The above-mentioned parameter adjustment is performed for each patch of the color patch chart. If a contradiction occurs, the following process is performed. (1) If a parameter changed when adjusting Idea1_patN = 1 (should be on-pixel) is changed when adjusting Ideal_patN = 0 (should be off-pixel) and a contradiction occurs, one of the following judgments should be made in advance. Allows you to select the conditions. (1a) Prioritize the adjustment of “Ideal_patN = 1”,
It does not change when adjusting N = 0. (1b) Prioritizing the adjustment of “Ideal_patN = 0”, the Ideal_pat
Update the parameter at N = 1.

(2) When a parameter changed at the time of adjusting Idea1_patN = 1 (should be on-pixel) is changed by another patch at the adjustment of Ideal_patN = 1 (should be on-pixel) to cause a contradiction, the patch is previously applied. Weighting is performed for each item, and whether to update the parameter is determined based on the weight.
That is, the result for the patch with the larger weight is emphasized.

(3) If a parameter changed at the time of adjusting Ideal_patN = 0 (should be off-pixel) is changed by another patch at the adjustment of Ideal_patN = 0 (should be off-pixel) to cause a contradiction, the patch is previously set. Weighting is performed for each item, and whether to update the parameter is determined based on the weight.
That is, the result for the patch with the larger weight is emphasized.

Although the size of the patch of the color patch chart is 3 × 3 pixels here, the patch of other size can be processed in the same manner. Further, the threshold value for determining the on / off of the patch may be set appropriately.

The dictionary adjustment can be similarly performed for the output system device. A case where the output device is a printer will be described with reference to FIG. The printer 38 inputs the electronic data of the color patch chart (S40) and prints the color patch chart on paper (S42). A color measuring device measures the color patch chart (S44) to obtain (multi-valued) color information of each patch (S46). The binary information of the ideal color patch chart obtained by the reference printer is stored in advance (S48), the color information is compared with the ideal binary information (S50), and the dictionary is corrected based on the result. Yes (S52). Although detailed description is omitted,
In this comparison, the color information is binarized using a dictionary as in the case of FIG. 3, and the result is compared with the ideal binary information. Thus, it is possible to accurately recognize that the image including the specific image has been printed.

[0032]

Since the content of the parameter (dictionary) defining the color space in which a specific image exists can be automatically adjusted,
The large amount of data required to set the optimal dictionary for each model is no longer required.

Anyone can make an adjustment by automating the adjustment work of the dictionary, which is performed only by an expert.

[Brief description of drawings]

FIG. 1 is a block diagram of an image recognition processing device of the present invention.

FIG. 2 is a diagram showing an example of a color range of a specific image.

FIG. 3 is a flowchart of dictionary adjustment for input devices.

[Figure 4] Color patch chart diagram

FIG. 5 is a flowchart of dictionary adjustment for an output device.

[Explanation of symbols]

10 CPU, 12 ROM, 14 RAM,
22 keyboard, 30 flexible disk, 34 hard disk, 38 printer,
42 Scanner.

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) H04N 1/46 H04N 1/40 ZF term (reference) 5B057 AA20 CA01 CA16 CB01 CB16 CE16 CH01 CH11 DB06 DC25 5C077 LL14 MM27 MP08 PP32 PQ12 PQ20 PQ22 RR02 5C079 HB01 LA08 LA34 MA01 MA10 MA11 NA27 5L096 AA02 AA06 BA20 FA15 HA07 MA03

Claims (5)

[Claims] 1. A step of storing a parameter that defines a color space in which a specific image exists in a storage means, a step of processing input image data using the parameter stored in the storage means, and a step of processing the input image data. Comparing the processed data with reference data for the input image data, and based on the result of the comparison, modifying the parameters so that the processed data of the input image data matches the reference data. , An image processing program executed by a computer. 2. The method according to claim 1, wherein the processing step, the comparing step and the correcting step are repeated until the processed data of the input image data matches the reference data. Image processing program. 3. The program according to claim 1, further comprising the step of inputting image data from an image input device that reads an image. 4. The program according to claim 1, further comprising a step of storing processed data of input image data in a storage means. 5. A parameter that defines a color space in which a specific image exists is stored in a storage unit, the input image data is processed using the parameter stored in the storage unit, and the processed data of the input image data is stored in the storage unit. An image processing method of comparing the input image data with reference data, and modifying the above-mentioned parameter based on the result of the comparison so that the processed data of the input image data matches the reference data.