JP5690668B2 - Automatic transaction device with image trail function - Google Patents

Description

  The present invention relates to a trail management system for an automatic transaction apparatus for financial use.

  In an automatic transaction apparatus for financial use such as ATM, there is a need to store an image of a banknote as evidence that a banknote that has been deposited / withdrawn is correctly transferred. This means that an image of the transaction of the banknote handled in the automatic transaction apparatus is left as an image, and when a trouble occurs during the transaction, the image of the banknote is presented and used as transaction evidence.

  Japanese Patent Application Laid-Open No. H10-228561 discloses a method for determining the authenticity of a bill by inserting a magnetic material. Patent Document 2 discloses a method for registering a paper piece including tag information of a paper piece input from a keyboard or the like in addition to the feature amount of the paper piece based on the transmission image in the paper piece discrimination method using transmitted light. These are techniques for compensating the uniqueness of banknotes, and are not intended for image trail management. Patent Document 3 describes a process for storing an image in an automatic transaction apparatus. This is a process for configuring the screen interface, and is different from banknote image trail storage.

Japanese Laid-Open Patent Publication No. 2007-164293 Japanese Laid-Open Patent Publication No. 2006-18525 JP-A 63-216169

  Since there is a limit to the capacity and calculation capacity of the storage device in the automatic transaction apparatus, and there is a band limitation when sending an image to the bank side, compression processing is required when saving / transferring the image. Therefore, for image capturing and compression performed in the automatic transaction device, it is necessary to leave an image with an image quality that can distinguish writing, dirt, scratches, etc., which are unique features of individual banknotes, and further, within the ATM device In order to perform compression, an algorithm for compressing an image with a low calculation amount and high image quality is required.

  In the image compression processing in the automatic transaction apparatus having the image trail function according to the present invention, the calculation ability and the storage capacity are limited by improving the compression efficiency and the processing speed by utilizing the feature of “image of banknote”. The purpose is to realize functions on an automatic transaction apparatus.

  In the banknote handling apparatus of the present invention, a deposit / withdrawal unit for depositing / withdrawing banknotes, a storage unit for storing banknotes, a discrimination unit for identifying banknotes, and an image of the banknotes when the banknotes pass through the discrimination unit. An image capturing unit; a transport mechanism that transports banknotes; a compression processing unit that compresses banknote images captured by the image capturing unit; and a storage unit that stores data of banknote images subjected to image compression processing by the image processing unit; It is characterized by having.

  Preferably, a discrimination part acquires banknote identification information which shows the specific characteristic for every banknote, memorize | stores banknote identification information in a storage part, and based on the banknote identification information memorize | stored in the storage part, an image imaging part is a banknote. It is determined whether to capture an image.

  In still another aspect of the present invention, the compression processing unit is based on identification information including at least one or more of information on the denomination, front and back, orientation, and banknote deterioration degree of the banknote acquired by the identification unit. The banknote image is compressed.

  Preferably, the compression processing unit performs edge detection processing and adaptive blur processing of the banknote image based on the discrimination information, and more preferably, the compression processing unit extracts the discrimination information from the sample image of the banknote stored in the storage unit. The sample image of the same denomination as that of the banknote image is selected, a difference image between the selected sample image and the banknote image is acquired, and the difference image is compressed.

  According to the present invention, the banknote status at the time of discrimination as evidence of banknote trading can be accurately stored. Further, according to still another aspect of the present invention, it is possible to realize an image with reduced block noise with a capacity lower than the JPEG compression size while leaving unique features such as banknote dirt, writing, and scratches. The amount of calculation can also be suppressed.

It is a block diagram which shows the structure of the automatic transaction apparatus system of this invention. It is an external view of the automatic transaction apparatus of this invention. It is a hardware block diagram of the present invention. It is a transition diagram of the hardware and bill processing of the present invention. It is a processing block diagram for obtaining an image trail in the present invention. It is a processing block diagram for performing image compression in the present invention. It is a flowchart of the banknote process in this invention. It is a figure which shows the structure and compression parameter of the sample dictionary in this invention.

  In the image compression processing in the automatic transaction apparatus having the image trail function according to the present invention, the calculation ability and the storage capacity are limited by improving the compression efficiency and the processing speed by utilizing the feature of “image of banknote”. Realize functions on automatic transaction equipment and eliminate the problem of image trail management. The following points are mentioned as points of the present invention.

  1) Cooperation linked compression: Depending on the result of banknote discrimination, the compression dictionary / algorithm is switched to execute the banknote compression process. As the discrimination information, the denomination, front / back, orientation, and deterioration degree (distance at the time of discrimination) of the banknote are passed to the image compression unit, thereby dynamically changing the dictionary compression process and the compression dictionary. The distance at the time of discrimination here refers to some scale used in determining the acceptance / rejection of the banknote and determining the type at the time of banknote discrimination. For example, there is a distance obtained by considering the darkness of a pixel acquired by scanning a specific position on the banknote surface with a sensor as a vector amount and calculating the Euclidean distance between this and the dictionary.

  2) Fluctuation difference dictionary: A fluctuation amount when a banknote is imaged is registered as a difference dictionary. In the compression process using variable blurring / partial difference, the calculation requiring a large amount of calculation is simplified and speeded up. By pseudo-moving the partial area and holding the DCT conversion data in the dictionary, it is possible to speed up the movement difference calculation process and perform high-precision collation.

  3) Image processing cooperation: The amount of calculation of image compression is reduced by reusing DCT conversion data once created for an input image in the subsequent image compression process.

  4) ATM device capable of confirming the trail of the image of the transaction banknote: taking the image at the identification unit, processing the image obtained by the identification unit for image compression, etc., the same banknote being identified multiple times In order to prevent double image processing because it passes, it has a function to prevent double processing by a mechanism for determining uniqueness of banknotes (banknote identification information) such as serial number recognition, sequence counter, paper pattern discrimination, etc. To do.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  First, the external appearance of the entire automatic transaction apparatus system to which the present invention is applied will be described. FIG. 1 shows a state in which a plurality of automatic teller machines 0101 are connected to a host computer 0100 via a network 0120. The guidance device 0102 installed in the automatic teller machine with a smaller usage amount displays a display content 0110 indicating that a transaction is possible. Other guidance devices 0102 display display content 0111 indicating that preparation is in progress. Thereby, it is possible to guide the customer to the automatic teller machine on which information indicating that the transaction is possible is displayed on the guidance device 0102. Thereby, for a financial institution, there is an advantage that the amount of cash stored in the ATM can be easily managed by making the usage amount of banknotes in ATM uniform.

  The external appearance of each automatic transaction apparatus 0101 is shown in FIG. Reference numeral 0200 denotes a case of an automatic transaction apparatus, 0201 denotes a passbook entry / exit, 0202 denotes an ID card entry / exit, 0203 denotes a deposit / withdrawal port, and 0204 denotes a display input device such as a touch panel type liquid crystal display. Reference numerals 0210 to 0217 denote sensor devices for supporting an interface between the automatic transaction apparatus and a human. For example, 0210 is a camera, 0211 is an iris sensor, 0212 is a microphone, 0213 is a retina sensor, 0214 is a fingerprint sensor, and 0215 is A palm print sensor, 0216 includes a finger vein sensor, and 0217 includes a palm vein sensor.

  A schematic diagram of a hardware configuration for realizing the automatic transaction apparatus 0101 is as shown in FIG. This embodiment of the present invention is performed by the apparatus described in FIG. FIG. 3 is a functional block diagram schematically illustrating the internal hardware configuration of the automatic transaction apparatus according to the present invention. The automatic transaction apparatus 0101 of this embodiment includes an operation terminal device 0302, a display terminal device 0303, an external storage device 0305, a memory 0306, a central processing unit 0307, a communication device 0309, and a communication line 0308 that connects these components to each other. In addition, it has an image pickup device 0301 for obtaining a trail of banknote transactions, and a 0304 device group for controlling peripheral modules and sensor groups such as banknote conveyance and discrimination. The configuration of the central processing unit, the memory, the external storage device, etc. of the automatic transaction apparatus 0101 may be an embedded microcomputer. The operation terminal device 0302 is, for example, a keyboard or a touch panel, and is used to input a user instruction or data to the automatic transaction device 0101. The display terminal device 0303 is a device that displays text, images, and the like, such as a liquid crystal display device. For example, a trail image acquired as a result of the banknote transaction can be presented to the user using this display device. The external storage device 0305 is a storage device such as a hard disk device or a flash memory, and stores a trail image, discrimination data, transaction information, and the like. The trail image is stored in combination with information such as transaction date and time, transaction user, banknote denomination and authenticity discrimination result. These pieces of information stored in the ATM are transmitted to the host side through the communication device 0309 as necessary.

  Furthermore, a program executed by the central processing unit 0307 to realize the present embodiment may be stored. The memory 0306 is, for example, a semiconductor memory, and stores a program executed by the central processing unit 0307, data to be referred to, and the like. At least a part of the program and data stored in the external storage device 0305 may be copied to the memory 0306 as necessary. The central processing unit 0307 executes the program stored in the memory 0306 and controls the operation terminal device 0302, the display terminal device 0303, the external storage device 0305, the peripheral device 0304, the image capturing device 0301, and the communication device 0309 as necessary. To do. In the following description, the processing executed by the automatic transaction apparatus 0101 having the image trail function is actually executed by the central processing unit 0307. The communication device 0309 is connected to a network and serves as an interface for communicating with other devices connected to the network.

  FIG. 4 shows a conceptual diagram of the hardware configuration shown in FIG. 3 in detail along with an example of an ATM device. For example, the central processing unit 0307, the memory 0306, the external storage device 0305, and the like may be provided in either or both of the identification device 0402 and the control device 0403 shown in FIG. This belongs to the design matter of the device.

  A block diagram of the automatic teller machine 0101 is shown in FIG. Deposit / withdrawal unit 0401, identification device 0402 for identifying the denomination / true / false of cash, communication device 0404 for connecting to a network, display terminal device 0405 such as a liquid crystal display, operation terminal such as a touch panel display A device 0406, a temporary stacker 0410 for primarily storing cash, a transport device 0411, a reflux box 0412 to 0414 capable of automatically storing and taking out cash, a non-reflux box 0415 capable of only automatically storing cash, and the above devices The automatic cash transaction apparatus 0101 is comprised from the control apparatus 0403 to control. The denomination here refers to how much money the banknote is in, and the inserted state such as the front and back of the banknote, upright / inverted, etc., authenticity means whether the banknote is genuine or counterfeit, Is a word which means whether the state of a banknote is normal or has deteriorated considerably.

  Explain deposit transactions. When a user of the automatic transaction apparatus inserts a bill into the automatic transaction apparatus 0101 from the insertion slot 0203 into the deposit / withdrawal unit 0401, first, the control device 0403 controls the transport device 0411 to transport the bill from the deposit / withdrawal unit 0401 to the identification device 0402. To do. Hereinafter, it will be described that the control device 0403 controls the transport device 0411 to transport the banknotes, simply that the control device 0403 transports the banknotes. The identification device 0402 performs denomination identification, authenticity identification, and correctness identification for each one of the input banknotes. The control device 0403 receives the identification result of each banknote identification for one banknote.

  There are roughly three processes executed by the identification device 0402. It is denomination identification, authenticity identification, and damage identification. These are basically processed in the same procedure. First, a sensor signal is acquired. Subsequently, feature extraction suitable for each discrimination is performed, and the feature vector obtained as a result is applied to a discriminator to obtain a discrimination result. Here, a system such as Mahalanobis distance, support vector machine, vector quantization, and three-layer perceptron is used for the discriminator. Eventually, a threshold process determines whether the denomination is denomination or does not apply to any denomination. Hereinafter, when the denomination can be determined, the denomination is determined, and when not, the denomination is not determined. When the denomination is determined, the front and back sides and orientation of the input bill are also determined. Further, a determination is made as to whether it is a genuine note if it is genuine or not, and a determination result whether it is true or not if it is correct or not. Furthermore, in general, a discrimination error value can be obtained from the discriminator. For example, when it is determined whether the bill is 1000 yen using the Mahalanobis distance, it means that the smaller the distance value, the more likely it is 1000 yen. In this case, the distance value corresponds to the identification error value. The damage determination is a process for identifying whether the banknote is clean enough to endure circulation or is a banknote to be collected due to dirt or damage.

  The control device 0403 receives the identification result of each banknote identification for each banknote, and controls the conveyance destination. That is, if the denomination is confirmed in the denomination identifying process, the banknote is sent to the temporary stacker 0410, otherwise the banknote is sent to the deposit / withdrawal unit 0401. If it is determined that the bill is genuine, the bill is sent to the temporary stacker 0410. Otherwise, the bill is sent to the deposit / withdrawal unit 0401. It is determined whether there are still banknotes to be processed. If there are still banknotes, the same processing is repeated. Through the above processing, banknotes other than banknotes whose acceptance has been rejected are stored in the temporary stacker 0410.

  At this time, when the user of the automatic transaction apparatus inputs a transaction reject, the banknotes stored in the temporary stacker 0410 are returned to the deposit / withdrawal unit 0401. When the transaction is continued, the bills are further conveyed from the temporary stacker to each box. That is, the control device 0403 starts processing one by one for the bills sent to the temporary stacker 0410. First, one banknote in the temporary stacker 0410 is conveyed to the identification device 0402. The identification device 0402 performs each identification by the above-described processing, and the control device 0403 receives the identification result. In the denomination identification, it is determined whether or not the denomination of the banknote is confirmed. When the denomination is confirmed, the authenticity identification process is executed. Otherwise, the banknote is stored in the non-reflux box 0415. It is determined whether or not the bill is a genuine note based on the result of authenticity identification. If it is a genuine note, a damage identification process is executed, and if not, the bill is stored in the non-reflux box 0415. It is determined whether or not the bill is a correct bill based on the result of the damage identification. If it is determined that the bill is not a correct bill, the bill is stored in the non-reflux box 0415. If the bill is a regular note, the control device 0403 stores the bill in the reflux boxes 0412 to 0414 according to the denomination identification result. It is determined whether or not there are still banknotes to be processed in the temporary stacker 0410. If there are more banknotes, these processes are repeated. The banknote image is captured, compressed, and stored in the identification device 0402 and the control device 0403 here.

  FIGS. 5 and 6 show how trail images are captured and stored in the present invention. FIG. 5 shows a functional block diagram of processing by cooperation between the identification device 0402 and the control device 0403. First, when the bill passes through the identification device 0402, the sensing unit 0501 acquires various sensor data and images of the bill. The sensor data shows the state of the banknote obtained from magnetism, light, etc., but here we will use enough sensor data to perform banknote discrimination (denomination judgment, authenticity judgment, damage judgment). do not do. The banknote image and sensor information obtained as a result are first sent to the banknote recognition unit 0502. In the banknote identification unit 0502, the above-described banknote identification (denomination identification, authenticity identification, and damage identification) is performed. The discrimination information 0504 combined with the denomination of the banknote, front and back, orientation, distance value at the time of banknote identification, and the banknote image are sent to the image compression unit 0503. Further, the banknote identification unit stores information (banknote identification information 0505) for determining the uniqueness of the banknote as a dictionary. The image compression unit 0503 switches the compression dictionary 0506 using the discrimination information 0504 when compressing the sent banknote image.

  The compression dictionary includes sample images and compression parameters. For example, sample images can be switched using denominations and front / back information. Moreover, compression efficiency can be improved by designating the blurring degree of adaptive blurring at a specific part of the banknote (for example, a pattern part at the lower right or a serial number part) depending on the direction of the banknote. Since the denomination and the front and back of the scanned image can be identified by the discrimination information, the sample image used for compression can be switched. Moreover, since the degree of deterioration of the state of the banknote is known by the damage determination, for example, by adjusting the parameters of the image blurring process, for high-quality banknotes, the blurring process is increased to increase the compression capacity. In order to leave the unique features clearly, there are processes such as reducing blurring at the expense of the compression capacity. In addition, since it is important to determine whether the banknote is a false bill or not by means of authenticity determination, there may be a means of performing almost no blurring process in order to leave a high-definition image. Details of the sample dictionary and compression parameters will be described later.

  In addition, the same banknote passes through the identification device 0402 several times. However, if banknote identification information that can confirm the uniqueness of each banknote is used, the same banknote is doubled at the time of image capturing in the sensing unit 0501. It is also possible to avoid the process of imaging. The banknote identification information is, for example, a banknote counter, data storing a paper pattern of banknotes, data storing a serial number of banknotes, or the uniqueness of banknotes is easily determined. Information can be used. As a result, it is possible to prevent double operation of image capturing and processing, and it is possible to reduce the amount of calculation. For example, the meaning of the uniqueness of banknotes includes banknote blurring, folding marks, dirt due to banknote rubbing, etc., and graffiti written with a pen or the like.

  FIG. 6 is a diagram showing an outline of image compression processing.

  The input banknote image 0601 is subjected to an edge detection process 0602 and an adaptive blur process 0603. For edge detection, for example, a Sobel filter used for edge extraction in the vertical and horizontal directions can be used. The adaptive blur process is a process of broadly blurring a thin pattern portion and leaving a sharp image. Details of the adaptive blur processing will be described later. When the adaptive blurring process 0603 and the edge detection process 0602 are performed, the parameters can be adjusted using the discrimination information 0613 obtained by the bill identifying unit 0502. Next, the color resolution is roughened by color resolution compression processing 0604. The color resolution compression processing means processing for reducing, for example, 256 gradations of each color in the RGB space to 128 gradations or 64 gradations. An RGB space, HSV space, YCbCr space, or the like can be used as the color space.

  Next, the image is divided into small region units of 8 × 8 pixels, and processing of two-dimensional discrete cosine transform and coefficient quantization (DCT transform 0605) is performed for each small region. This is for expressing a pattern of a small region of 8 × 8 pixels by converting it to a spatial frequency, and is a method generally used in JPEG compression or the like. As a result, the amount of information required for expressing the pattern can be greatly reduced.

  Next, a difference between the image serving as a sample of the banknote and the input image is obtained. This is the optimal difference calculation process 0606, which uses the discrimination information 0613 and uses the denomination / front and back sample dictionary corresponding to the input image. However, misalignment occurs in the banknote depending on the situation at the time of imaging and the state of the banknote. The variation difference dictionary 0607 records the result of DCT transform of a small region unit of 8 × 8 pixels in consideration of the positional deviation of the sample image as a set of vector data. The variation difference dictionary 0607 is created by substantially the same process 0612 performed on the input image. In the optimum difference calculation 0606, for each 8 × 8 pixel small area on the input image, a small area of the sample image that best matches the small area is obtained. As a result, a partial movement amount 0608 of the dictionary image and a movement difference image 0609 that is a difference between the sample image and the input image are obtained. The partial movement amount means a position found by dividing the dictionary image into several small areas, moving each small area, and searching for a place that best matches the input image. Data compression processing is performed on these data.

  In the data compression, the partial movement amount 0608 can be compressed by Huffman coding or arithmetic coding of the data string at the movement position, and the moving difference image 0609 can be applied with image compression processing such as JPEG. When a JPEG-based technique is used for image compression, the DCT conversion data created in process 0605 can be passed through the 0611 path.

  The meaning of performing such compression processing will be described. The paper ticket image can be considered to be composed of elements such as a common pattern, information (serial number) unique to the paper ticket, deterioration, and the like. Actually, the pattern part slightly varies due to scanning, printing position deviation, and paper expansion / contraction. Therefore, if data compression is expressed by a function H, H (pattern common to paper tickets) + H (pattern variation information) + H (Information unique to a paper ticket) = H1 + H2 + H3 can be decomposed, and if only H2 + H3 is targeted for compression, the capacity can be reduced. Moreover, since the common pattern H1 is preserve | saved as a dictionary image of a sample, it can be disregarded as the compression capacity | capacitance of an individual banknote. A difference image is used as means for decomposing information.

  In other words, the pattern H1 common to the paper ticket is registered in the dictionary, and the pattern variation information H2 and the unique image information H3 are left. The use of the difference image is a technique that is also used in moving picture compression such as MPEG, and is called motion compensation (MC). In general, in moving image compression, an object (or partial area) moving from the previous frame is detected in units of 16 × 16 pixel micro blocks, and the movement amount and the difference amount are stored. The collation of 16 × 16 micro blocks is performed based on a relatively simple scale with a small amount of calculation, such as using the absolute value sum of RGB as a reference. Furthermore, data compression of microblocks is performed by two-dimensional discrete cosine transform and quantization (DCT). However, it is not preferable in terms of processing time to apply the moving image compression process as it is to the banknote image. Moreover, since a banknote contains many fine patterns, it is not desirable also in terms of capacity to store this as it is.

  A process for preserving a fine pattern by blurring and storing banknote-specific dirt and writing is an adaptive blur process 0603. The adaptive blur process will be described. In adaptive blurring, the Gaussian distribution is changed in multiple steps based on edge strength. The Gaussian filter has a fixed size of 5 × 5 and is applied by adjusting the dispersion parameter σ stepwise, leaving the edge portion blurred. However, depending on the elements that make up the banknote, there is a pattern in which sufficient sharpness cannot be maintained just by using the edge strength as it is. In this case, as a dictionary, processing such as limiting the range to be blurred by adaptive blur at a place where a banknote is present is necessary. For this purpose, it is essential to use identification information.

A process for reducing the image capacity using the sample image of the banknote is a movement difference process (optimum difference calculation process 0606). Optimal matching, that is, calculation of small images with similar patterns, tends to be more stable than the RGB norm distance when the DCT minimization criterion is used. Another reason for the DCT minimization criteria is that later data compression is advantageous. In the difference calculation result of the two images based on the RGB standard (1-norm distance) and the DCT standard, a place where there is no difference between the two images is gray. Colors other than gray represent estimation errors. The larger the gray area, the more advantageous for later image compression. That is, the gray area means that the image information can be described by the pattern H1 common to the sample image = banknote. The problem of calculating the amount of movement so as to minimize the difference between the first to Nth components of the DCT and collating the range so that ± M pixels can be moved is to move the sample image in various ways. The obtained data can be expanded in advance as a set of DCT vectors and replaced with a problem of calculating a minimum distance with these vector sets. This makes it possible to increase the processing speed by utilizing packed arithmetic in an embedded processor or the like. In general, the compression used in moving images is mainly based on the detection of the movement amount by the RGB norm with respect to the previous frame. However, in the case of a banknote image, the sample image can be registered in the dictionary in advance, so the DCT conversion dictionary of the sample image can be stored. The advantage is that it can be prepared in advance.
Next, a processing flow of banknotes stored in the temporary stacker 0410 or the storage box groups 0412 to 0415 will be described with reference to FIG. First, one banknote in the temporary stacker or the storage box group is conveyed to the identification device 0402 (0701). The identification device 0402 performs sensing, imaging, and identification processing, and the control device 0403 receives the identification result (0702). In the denomination identification, it is determined whether or not the denomination is confirmed (0703). If it is confirmed, the process proceeds to 0704. Otherwise, the banknote is stored in the non-reflux box 0415 (or deposit / withdrawal unit 0401) (0707). In 0704, based on the authenticity identification result, it is determined whether or not the bill is a genuine note. If it is a genuine note, the processing proceeds to 0705. (0707). In 0705, it is determined whether or not the bill is a correct note based on the result of the damage identification. If it is a correct note, the flow proceeds to 0706. Otherwise, the bill is stored in the non-reflux box 0415 (0707). The control device 0403 stores banknotes in the reflux boxes 0412 to 0414 according to the denomination identification result of 0706 according to the denomination. It is determined whether or not there is still a bill to be processed in the temporary stacker 0410 (0708). If there is still a bill, the process returns to 0701, and if not, the process proceeds to 0709. In 0709, each banknote image and discrimination information in all the banknotes traded are received from the identification device 0402, and compression processing and storage processing of the image are performed.

  The object to be received may be either the identification device 0402 itself, the control device 0403, or both. This is due to the design guidelines of the ATM device itself. For example, if the CPU held by the control device is powerful, it is advantageous to perform image compression here. On the other hand, when the discrimination apparatus has a special CPU for image processing, it is advantageous to perform image compression here. When the trail information is obtained in the host computer or the operation terminal device, the compressed image information is transmitted. In addition, banknote identification information is used so that the identification device 0402 does not capture a banknote image twice. The image information is transmitted to, for example, a host computer of a financial institution behind the ATM device that performs financial transaction processing. The transmission of the image information is performed through the communication device 0404.

  Finally, the structure of the sample dictionary and details of the compression parameters are described. In the dictionary, sample image data 0801 and compression parameter information 0802 are recorded. In the sample image data, information 0803 such as the denomination / front / back / upside-down of the image recorded in the dictionary, information 0804 such as the size of the banknote image and the number of divisions of the small area, etc. A plurality of DCT data 0805, which is image data for each small area, are recorded. In the small area DCT data, a number 0806 for uniquely identifying the small area, a movement amount 0807 when the small area is intentionally shifted, and the DCT data at that time are recorded as a vector quantity 0808. Therefore, when an input image is input, the input image is divided into small regions and subjected to DCT transform. Since the image data of the small region considering movement is recorded as DCT vector data in the sample dictionary, this 2 Image matching corresponding to the deviation of the banknote image can be performed with a relatively small amount of calculation only by performing vector matching between the two.

  Further, as parameter data, information 0810 describing an image processing procedure, color space information to be used, and the like, a blur parameter 0811 suitable for a low-quality to high-quality banknote image, an edge extraction parameter 0812, a movement difference parameter 0813 There is. For example, the blur parameter is a list that defines the width of the Gaussian blur function for each image quality. The edge extraction parameter is a list that defines window widths and image filter values for calculating edge information for each image quality. The moving difference parameter includes a weighting coefficient used for calculating a weighted Euclidean distance when calculating an optimal collation. Moreover, if the component parameter 0814 that a serial number exists in the specific position on a banknote is used, the quality of compression can be controlled for every component.

0101 ... Automatic cash transaction device 0120 ... Network 0100 ... Host computer 0201 ... Passbook entry / exit port 2022 ... ID card entry / exit port 0203 ... Deposit / withdrawal port 0204 ... Display input device 0210 0217 ... Sensor device 0302 ... Operation terminal device 0303 ... Display terminal device 0305 ... External storage device 0306 ... Memory 0307 ... Central processing unit 0309 ... Communication device

Claims (12)

A deposit / withdrawal unit for depositing / withdrawing banknotes;
A storage unit for storing banknotes;
A discrimination part for discriminating bills;
An image capturing unit that captures an image of the banknote when the banknote passes through the discrimination unit;
A transport mechanism for transporting banknotes;
A compression processing unit that performs compression processing of the banknote image captured by the image capturing unit;
Have a, a storage unit for storing the data of the image compression processing bill image by said image processing unit,
The banknote handling apparatus, wherein the compression processing unit performs the compression processing by switching a method of compression processing of the banknote image based on banknote discrimination information acquired by the discrimination unit. The bill handling apparatus according to claim 1,
The said discrimination part acquires the banknote identification information which shows the specific characteristic for every banknote,
The storage unit stores the banknote identification information acquired by the discrimination unit. The banknote handling apparatus according to claim 2,
The said image imaging part images the image of a banknote based on the said banknote identification information stored in the said storage part, The banknote handling apparatus characterized by the above-mentioned. The bill handling apparatus according to claim 1,
The said discrimination information is information containing at least 1 or more among the information regarding the money type of a banknote, front and back, direction, and the deterioration degree of a banknote, The banknote handling apparatus characterized by the above-mentioned. The bill handling apparatus according to claim 1,
The banknote handling apparatus, wherein the compression processing unit performs an edge detection process and an adaptive blurring process on the banknote image based on the discrimination information. The banknote handling apparatus according to claim 5,
The said banknote handling apparatus determines the range which performs the adaptive blurring process of the said banknote image based on the said discrimination information. The bill handling apparatus according to claim 1,
The compression processing unit selects a sample image of the same denomination as the banknote image from the sample image of the banknote stored in the storage unit based on the discrimination information, and the selected sample image and the banknote A banknote handling apparatus that obtains a difference image from an image and performs compression processing of the difference image. It is a banknote handling apparatus of Claim 7, Comprising:
A banknote handling apparatus that divides the banknote image into a plurality of areas, collates with the sample image for each area by vector calculation, and acquires the difference image. A deposit / withdrawal unit for depositing / withdrawing banknotes, a storage unit for storing banknotes, a discrimination unit for identifying banknotes, an image capturing unit for capturing images of banknotes, a compression processing unit for compressing images of banknotes, and banknotes A program for controlling a bill handling device comprising a transport mechanism for transporting,
The transport mechanism transports the banknotes deposited by the deposit and withdrawal unit;
A step in which the image capturing unit captures an image of a banknote when the banknote passes through the discrimination unit;
Causing the compression processing unit to perform the compression processing by switching the method of compression processing of the banknote image based on the banknote discrimination information acquired by the discrimination unit. A program characterized by The program according to claim 9, wherein
Acquiring the banknote identification information indicating the unique characteristics of each banknote,
Storing the banknote identification information in a storage unit;
And a step of determining whether or not the image capturing unit captures an image of a banknote based on the banknote identification information stored in the storage unit. The program according to claim 10,
The said discrimination information is information containing at least 1 or more among the information regarding the money type of a banknote, front and back, direction, and the deterioration degree of a banknote, The program characterized by the above-mentioned. The program according to claim 10,
The compression processing unit executes a step of performing edge detection processing and adaptive blurring processing of the banknote image based on the discrimination information.

Images