JP3220178B2 - Image processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly to an image processing apparatus having an image recognition function for an object to be copied.

[0002]

2. Description of the Related Art Conventional black-and-white copiers and color copiers have been developed with particular importance as to how faithfully the original image itself can be copied.

[0003]

However, since reproduction of an image is simply performed faithfully by copying, there are the following drawbacks. That is, (1) a disadvantage that an original that can not be copied, for example, a bill such as a bill or a security, can be easily copied because anything can be copied; Another disadvantage is that, under normal conditions, a copy for which copying is not permitted is easily copied.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and enables control of output to an image having a predetermined symbol, and appropriately controls copying by detecting the presence of specific information. It is an object of the present invention to provide a possible image processing device and image processing method. For example, an object of the present invention is to provide an image processing apparatus capable of controlling an output method according to the type of a symbol included in a document. It is another object of the present invention to provide an image processing apparatus capable of accurately detecting a symbol in a document image and controlling an output method even if the symbol is slightly deformed.

[0005]

In order to achieve the above-mentioned object, an image processing apparatus according to the present invention comprises: input means for inputting image data corresponding to a document image in which predetermined invisible symbols are embedded; Recognition means for recognizing the invisible symbol from the image data input by the input means and determining the type thereof, and a method of outputting the image data input by the input means corresponding to the type of the symbol recognized by the recognition means. It is characterized by comprising a deciding means for deciding, and an output means for performing output processing according to the output method decided by the deciding means. An input unit for inputting image data corresponding to a document image including a predetermined invisible symbol on a line-by-line basis, and a back-propagation method for each bit of one line of image data input by the input unit. An operation is performed using the weight information corresponding to each bit that has been performed, and an analysis unit that outputs a standardized signal, and analyzes based on a plurality of lines of standardized signals output from the operation unit. Recognizing means having a neural network for recognizing the invisible symbols and determining the type thereof; determining means for determining an output method of image data input by the input means according to a recognition result of the recognizing means; Output means for performing output processing in accordance with the output method determined by the means.

[0006]

According to this structure, the input means inputs image data corresponding to the original image in which the predetermined invisible symbols are embedded, and the recognition means recognizes the invisible symbols from the image data input by the input means. And determining the type of the image data. The determining means determines an output method of the image data input by the input means in accordance with the type of the symbol recognized by the recognition means, and the output means determines the output method determined by the determining means. The output processing is performed according to. According to another configuration, the input unit inputs image data corresponding to the original image including the invisible predetermined symbol in line units, and the arithmetic unit inputs each bit of the one-line image data input by the input unit. For each time, a standardized signal is output by performing an operation using the weight information corresponding to each bit determined in advance by the back propagation method, and a recognition unit having a neural network is output from the operation unit. An analysis is performed based on the standardized signals for a plurality of lines to recognize the invisible symbol and determine its type, and the determining unit outputs the image data input by the input unit according to the recognition result of the recognition unit. The output means performs an output process according to the output method determined by the determination means.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment according to the present invention will be described below in detail with reference to the accompanying drawings. A copying machine will be described below as an application example of the present invention in the embodiment, but the present invention is not limited to this, and can be applied to various devices such as a printer and a communication device. Further, in the following description, a document such as a bill, a gift certificate, a security, and the like will be referred to as a specific document.

<First Embodiment> FIG. 1 is a diagram for explaining a forgery prevention method according to a first embodiment of the present invention. In the figure, (a), (b) and (c) show three different forgery prevention symbols, respectively. The forgery prevention referred to here includes not only copying of the specific manuscript, but also manuscripts such as company gazettes and personal items. As shown in the figure, each document has areas α1 to α3 and β1 to β3 having different reflectances, and one of the areas α1 to α3 indicates a symbol. By determining the shape of each of the areas α1 to α3, forgery prevention symbols are extracted.
In the case of (a), a symbol having a “ten” shape,
In the case of (b), a symbol having the shape of “X”,
In the case of (c), symbols each having a shape of “” ”are shown.

On each document, an image of a forgery-prevented object (for example, a copyrighted object) is actually printed. The copier of this embodiment scans the entire screen from the beginning with a line sensor. For the symbol on each document, for example, the reflectance of the infrared light region is changed. By using a sensor (such as a CCD sensor) of a type that is sensitive to signals of wavelengths in the infrared region, even if the symbols cannot be determined by visual inspection, the above-mentioned sensor scans the original and scans them. It is possible to obtain an electric signal including information on the difference between the reflectances of the two regions.

FIG. 2 is a block diagram showing the configuration of a first embodiment to which the image processing apparatus according to the present invention is applied. In FIG. 1, reference numeral 1 denotes a document, 2 denotes an n-bit line sensor, and 7 denotes a serial signal input type neuro sensor, which has its own input connected to a serial signal output terminal of the line sensor 2. Reference numeral 3 denotes a sample hold (hereinafter, referred to as "S / H") circuit, which connects the output terminal of the neuro element 7 to its own input terminal. 4 is a neural network, S
/ H circuit 3 has its own m input terminals connected to the m hold signal output terminals. Reference numeral 5 denotes a CPU for controlling the entire apparatus, and connects three output terminals of the neural network 4 to its own input terminal. Reference numeral 6 denotes a light source including a spectrum in the infrared region irradiated on the paper surface. Reference numeral 8 denotes a control signal bus and a clock line, which are a line sensor 2, an S / H circuit 3, a neural network 4, and a CP.
U5 is connected to the control terminal of the neuro element 7. 9
Denotes an alarm device, 10 denotes a power control device, and 11 denotes a printer device. Reference numerals 12, 13, and 14 denote signal lines for transmitting signals between the alarm device 9, the power supply control device 10, the printer device 11, and the CPU 5.

Next, the operation of the first embodiment will be described.

FIG. 5 is a flowchart for explaining a copy operation for preventing forgery according to the first embodiment.

The copying machine scans the original 1 using two line sensors, and the line sensor 2 converts the amount of light reflected from the light source 6 on the original 1 into an electric signal (step S).
1). The converted electric signal is a serial signal in which a signal for one line is decomposed into n bits and output from the line sensor 2. The neuro element 7 takes a product of each bit signal and weight information corresponding to the bit, and A signal obtained by normalizing the sum of n-bit signal products by a sigmoid function is output to its output terminal (step S2). The standardized output signals for each line are sequentially sampled and held by the S / H circuit 3 independently, and the m standardized signals obtained after the line sensor scans the entire original surface with m lines are obtained. When sample and hold by S / H circuit 3 for each line,
The neural network 4 analyzes the infrared reflectance from the signal and outputs the analysis result to three output terminals (step S3). The CPU 5 reads the determination result from the input port, and determines the meaning of the information on the document (step S4). The control of the entire signal is performed by the CPU 5.
Are executed through the signal line 8.

FIG. 3 is a diagram showing the contents of the neural network 4 according to the first embodiment. As shown in FIG. 3, the neural network 4 has a three-layer structure of an input layer, an intermediate layer, and an output layer, and is configured by connecting 2 × m + 3 neuron elements, and weight information is previously determined by a back propagation method. Has been determined. Here, the analysis of infrared reflectance is performed, and the method follows the description of FIG.

FIG. 4 shows a neuro element 7 according to the first embodiment.
It is a figure explaining the specific content of. As shown in the figure, in the neuro element 7, the serial signals a1 to an are sequentially input, and using the weight information Wn from the CPU 5, the calculation is performed by the sigmoid function shown in the following equation (1) and normalized. Output a signal. Equation (1) is

[0016]

(Equation 1) It is.

The CPU 5 includes an alarm device 9, a power control device 1
0, the signal lines 12 and 1
When the meaning of the information on the document 1 is determined, different commands are given to the warning device 9, the power supply control device 10, and the printer device 11 depending on the meaning of the symbol. For example, if the meaning of the symbol is the copy prohibition information, that is, if the original is a forgery prevention target, CP
U5 generates an alarm sound by the alarm device 9 based on the information, gives a caution, and stops the operation (step S5). If the meaning of the symbol is not copy prohibition information, the copying operation is continued (step S6).

If the alarm device 9 is not provided, the developing bias voltage of the power supply control device 10 is controlled so that the copied image is output as a blank sheet or as a solid black image. Thus, forgery copying can be prevented. In particular, the selection of solid black or blank is set by the administrator who performs maintenance and the like.

As described above, according to the first embodiment, a symbol indicating that the object is a forgery prevention object is given in the image or on the document,
By detecting the symbol at the time of copying, it is possible to easily determine whether or not the image to be copied or the original contains an image to be forged, and each apparatus can be determined according to the determination result. By changing the control method (for example, by sounding an alarm device to stop copying), forgery can be prevented by preventing copying. In particular, by using a neural network for the determination, the symbol can be easily determined.

<Second Embodiment> FIG. 7 is a block diagram showing the configuration of a second embodiment to which the image processing apparatus according to the present invention is applied. The description of the same configuration as that of FIG. 1 is omitted,
Only different configurations will be described. In FIG. 7, 38
Reference numerals 43 to 45 to 47, 49, and 50 indicate signal lines, respectively. Numeral 30 denotes an image correction circuit which performs shooting and other corrections on data read by scanning the information on the document surface line by line with the n-bit line sensor 2, and performs a neuro operation through a signal line 32 on the corrected data. The signal is sent to the circuit 31 and the image processing circuit 33. Reference numeral 31 denotes a neuro operation circuit, which includes the neuro element 7, S /
It has the same configuration as the H circuit 3, the neural network 4, and the CPU 5. Reference numeral 33 denotes an image processing circuit. In the case of a color copier, RGB read signals are Y, M,
The signal is converted into a signal by C, and the image is corrected and transmitted to the frame memory 34.

A print control circuit 35 controls the printer 37 according to an instruction from the neuro operation circuit 31. A printer 37 receives YMCK data from the frame memory 34 under the control of a print control circuit 35 and performs a printing operation. A color data control circuit 36 receives information on the frame memory 34 via a signal line 43 and feeds it back to the image processing circuit 33. A symbol information memory 44 holds symbol information. Reference numeral 34 denotes a frame memory for storing image data and symbol information in a superimposed manner.

Reference numeral 48 denotes an operation panel, which includes keys for instructing copy start and the like and setting various parameters, and a display unit for performing display corresponding to various operations.

Next, the operation of the above configuration will be described.

Image information on the document surface read by the line sensor 2 is normalized by shooting correction or the like by an image correction circuit 30 and then sent out to a signal line 32 to be sent to a neuro operation circuit 31 and an image processing circuit 33. Is entered. The neuro operation circuit 31 is realized by a configuration such as an array processor in which each neuro is constituted by a CPU having a risk structure, and the signal lines 12, 38, 3 according to the internal judgment result.
9 is sent a control signal. By that time, the data output on the signal line 32 is input to the image processing circuit 33, and the RGB signals are converted to the outline correction and the like and the RGB YMC
Is performed, output to the signal line 40, and stored in the frame memory 34 as printout data.

When the operation result of the neuro operation circuit 31 and the processed image indicate the prevention of forgery of an image relating to color, the color data control circuit 36 is switched to connect the signal lines 46 and 45 to connect the frame memory. The signal on the line 34 is returned to the image processing circuit 33, the color balance is changed, and a process of writing the frame memory 34 through the signal line 40 anew is performed.

Here, the neuro operation circuit 31 executes the same processing at a high speed in the CPU core array using the risks and the like of the neuron element 7, the S / H circuit 3, the neural network 4, and the CPU 5 in the first embodiment. The printer control circuit 3 is more
5 and a control information generation function of the color data control circuit 36 are also added.

In addition to the method of the first embodiment, in the case of color information, a method of characterizing the symbol distribution may be considered for the symbol determination.

When the neuro-operation circuit 31 detects a copy-inhibited symbol, the signal lines 39, 36, 4
The information of the symbol is displayed on the display unit (not shown) on the operation panel 48 through 7, 44, and 49, and at the same time, depending on the degree of the symbol, for example, if a bill is detected, the police At the same time as the alarm is sounded at the office 100, the copying operation is stopped according to the degree of the symbol. Of course, the alarm device 9 may sound an alarm.

It is assumed that the weight information is previously determined in the neural network by the back propagation method.

The neuro operation circuit 31 may be equivalent to the neuro element 7, the S / H circuit 3, and the neural network 4, and can be executed by a single super high speed RISC machine. is there.

The neuro operation circuit 31 is configured so that the information for each line can be reconstructed by using different sets of weight information, and a learning function that can easily extract similar deformation such as rotation reduction. It has.

Further, if the feature cannot be extracted by the same algorithm as in the first embodiment, an operation, for example, a rotation operation or a Fourier transform of colors, is performed on the data in the frame memory to generate different information. It is also possible to execute a new neural operation based on that, and execute a learning by a back propagation method so that feature extraction can be performed. In particular, weighting information is prepared in advance so that feature extraction of a topological image is easy.

Further, according to the determined symbol, the symbol is further changed, for example, information on the number of grandchild copies is changed, the status of the person making the copy is changed from a key, or an IC By inputting using a card or the like, the symbol can be changed, or a symbol indicating that a new copy image has been copied can be added, thereby facilitating forgery prevention. Further, there is an effect that the color balance can be easily changed according to the determined symbol. Further, by disturbing the color balance in accordance with the determined symbols and outputting the original image in a form that does not accurately output, it is easy to prevent forgery relating to colors. Furthermore, a function to add a symbol to prohibit copying in accordance with the detected image information or to detect a prohibited symbol and notify a function of an illegal copy to an external organization at the same time is provided. This is displayed on the display section of the operation panel, and copying is stopped, thereby facilitating prevention of forgery. FIG. 6 shows an example in which confidential information is inserted in the original image in advance by using the same method as the symbol data. At that time, the confidential information is detected only by a predetermined combination of weights of a predetermined neural network. The detected information can be output as output information according to the ID number of the memory card, and it is possible to prevent the information from being spread to an unspecified large number.

Further, according to the present invention, the symbol of the work is changed by changing the symbol of the work in accordance with the detected symbol information or the intention of the person making the copy, or by adding another forgery prevention symbol. The determination and the like may be facilitated.

Furthermore, in the present invention, only one copy-permitted symbol may be given. In this case, a symbol that protects the right of a specific individual regarding the color when detecting the one-time copy-permitted symbol and the color is detected. In such a case, the information of the image may be out of balance in the second and subsequent times, and a function of preventing the original image from being accurately copied may be provided to prevent the image from being unnecessarily spread.

[0036]

As described above, according to the present invention, an image processing apparatus capable of controlling output to an image having a predetermined symbol and suitably controlling copying by detecting the presence of specific information. Can be provided, and the control of image formation and copying can be flexibly performed. Specifically, it is possible to control the output method according to the type of the symbol included in the document. Also,
Even if the symbol in the original image is slightly deformed, it is possible to detect the symbol with high accuracy and control the output method.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a forgery prevention method according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a first embodiment to which the image processing apparatus according to the present invention is applied.

FIG. 3 shows a neural network 4 according to the first embodiment.
It is a figure which shows the structure of the contents.

FIG. 4 is a diagram illustrating specific contents of a neuro element 7 according to the first embodiment.

FIG. 5 is a flowchart illustrating a copy operation for preventing forgery according to the first embodiment;

FIG. 6 is a diagram illustrating a forgery prevention method according to a second embodiment.

FIG. 7 is a block diagram showing a configuration of a second embodiment to which the image processing apparatus according to the present invention is applied.

[Explanation of symbols]

 Reference Signs List 1 original 2 line sensor 3 S / H circuit 4 neural network 5 CPU 6 light source 31 neuro operation circuit 30 image correction circuit 33 image processing circuit 34 frame memory 44 symbol information memory 48 operation panel 11, 37 printer 9 alarm device

Continuation of the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 1/40-1/60 G03G 21/00-21/04

Claims (5)

(57) [Claims] 1. An input means for inputting image data corresponding to a document image in which a predetermined invisible symbol is embedded, and recognizing the invisible symbol from the image data input by the input means, and determining a type of the invisible symbol. Recognizing means, determining means for determining an output method of the image data input by the input means in accordance with the type of symbol recognized by the recognizing means, and output means for performing output processing in accordance with the output method determined by the determining means An image processing apparatus, comprising: 2. The image processing apparatus according to claim 1, wherein the output unit includes an adding unit that adds a new symbol or visible information to the image data according to a type of the symbol determined by the recognizing unit. Image processing device. 3. The image according to claim 1, wherein said output means has a notifying means for notifying that the type of symbol determined by said recognizing means is a symbol indicating forgery prevention. Processing equipment. 4. An input means for inputting image data corresponding to a document image including a predetermined invisible symbol on a line-by-line basis, and back-projecting each bit of the one-line image data input by said input means in advance. Calculating means for performing a calculation using the weight information corresponding to each bit determined by the method and outputting a standardized signal; and analyzing based on a plurality of lines of standardized signals output from the calculating means. To recognize the invisible symbol
And, recognizing means and a determining means for determining an output method of the image data input by said input means in accordance with the recognition result of said recognition means, the output method determined by said determining means having a neural network to determine the type An image processing apparatus, comprising: output means for performing output processing according to: 5. recognizes a reading step reading invisibly predetermined symbol reading the image information of the original embedded device, the invisible <br/> symbols from the image information read by the reading step, the a recognition step of determining the type, the corresponding type of the symbol recognized by said recognition step
Determine the output method of image information read in the reading process
When a specific type of symbol is recognized in the determination step and the recognition step,
Copy according to the output method determined in the determination step
A second symbol indicating that the image is
An image processing method, comprising: a copying step of performing copying with a printer in addition to information .