JP3674301B2 - Image processing method for carrier medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method for a conveyance medium, and in particular, a conveyance medium that takes out an image of a rectangular medium captured in a tilted state during conveyance from an image obtained by scanning while conveying the rectangular medium. The present invention relates to an image processing method.
[0002]
[Prior art]
For example, in an apparatus that discriminates banknotes, a denomination and authenticity determination are performed by reading an image of a banknote by an image sensor arranged in the middle of a conveyance path while conveying the banknote, and processing the read image. Yes. That is, a characteristic pattern portion such as a printed pattern or a watermark that is different for each denomination arranged at a predetermined position of the banknote with respect to the read image is extracted, and the characteristic pattern portion of the banknote serving as a reference is extracted from The banknotes are identified by comparison. Therefore, in order to accurately determine the denomination or the like, the coordinate position of the characteristic pattern portion to be extracted from the captured image of the banknote must be accurately known.
[0003]
By the way, although imaging of a banknote is performed while transporting a banknote, the width of the transport path is set in accordance with the banknote having the largest dimension among banknotes to be identified. Therefore, it is easy to bend in the conveyance direction during conveyance, and the sheet is conveyed in a skew state. A banknote image captured in such a skew state cannot accurately extract a feature amount. In view of this, normally, before actually performing banknote discrimination, skew correction of a banknote is detected from a captured image, and skew correction is performed to make the banknote image straight.
[0004]
For example, in the technique described in JP-A-8-212417, first, the entire banknote is imaged, and the outline of the banknote, that is, information on the short side and the long side of the banknote is obtained from the image data of the captured banknote. Thus, the skew angle of the banknote is obtained. Thereafter, the skewed image is rotated by the skew amount with the corner of the image as the origin, so that the image is straight.
[0005]
Next, discrimination processing is performed based on the skew-corrected image, and the banknote transport path is switched according to the discrimination result. That is, when it is determined that the banknote is correct by the discrimination process, the transport path is switched to the transport path toward the storage for storing the banknote, and it is determined that the banknote is not correct or cannot be discriminated. The transfer path is switched to the transfer path for returning such bills.
[0006]
As described above, in the case of banknote discrimination, first, the captured image is skew-corrected by preprocessing, the skew-corrected banknote image is subjected to discrimination processing, and the conveyance path is switched based on the discrimination result. ing. Therefore, this discrimination process must be completed before the conveyance path is switched.
[0007]
[Problems to be solved by the invention]
By the way, as counterfeit bill technology becomes more sophisticated, the algorithm of discrimination logic has become more complex, and the discrimination process has become more complicated accordingly. For example, two types of light sources having different emission wavelengths are used to illuminate a banknote when the banknote is imaged, and two types of banknote images are acquired by switching the light source for each scan line. Attempts have been made to improve discrimination accuracy by performing discrimination processing. In such a case, the time for the discrimination process is almost twice that of the normal discrimination process. Also in this case, of course, the discrimination process must be completed before the conveyance path is switched. Thus, in order to improve the discrimination accuracy, the processing time tends to become longer in the future, but the discrimination processing is performed while maintaining the conveyance speed of the banknote and the length of the conveyance path until the conveyance path is switched. There is a problem that it is difficult to lengthen only the time.
[0008]
The present invention has been made in view of these points, and shortens the pre-discrimination processing time, thereby increasing the time that can be spent on the discrimination processing, and adopting a complicated discrimination process. An object of the present invention is to provide an image processing method for a possible carrier medium.
[0009]
[Means for Solving the Problems]
In the present invention, in order to solve the above-mentioned problem, in the image processing method for a transport medium that corrects the inclination state of an image obtained by capturing an image while transporting the rectangular medium, the rectangular medium on which the medium detection sensor has been transported After the leading edge of the rectangular medium is detected, imaging for each scan line of the rectangular medium is started after a predetermined time determined by the conveying speed and the distance to the image sensor, and the medium detection sensor detects the trailing edge of the rectangular medium in the conveying direction. After detection, the indeterminate area arrival time until the imaging reaches the angle calculation indeterminate area before the trailing edge of the rectangular medium in the transport direction is calculated. Then, while correcting the angle of the tilted image, the imaging of the remaining angle calculation indefinite region and the angle correction of the tilted image are performed. A method is provided.
[0010]
According to such a conveyance medium image processing method, after the leading end of the rectangular medium is detected by the medium detection sensor, images are sequentially picked up from the leading end of the rectangular medium, and the conveyance speed is detected when the trailing end of the rectangular medium is detected by the medium detection sensor. Then, the length of the rectangular medium is calculated from the medium passing time, and the time until the imaging reaches the predetermined rectangular medium angle calculation indefinite region is calculated from the length of the rectangular medium. After that time has passed, when the imaging reaches the angle calculation indefinite region of the rectangular medium, the inclination angle of the medium image is obtained from the image data obtained so far, and the medium image is angle-corrected with the inclination angle. To do. In parallel with this, the remaining angle calculation indefinite region of the rectangular medium is imaged, and angle correction is performed with the previously obtained inclination angle. As a result, the angle correction has been completed in most parts of the medium image when the imaging of the rectangular medium is completed.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings, taking as an example a case where the present invention is applied to an image reading apparatus for banknote discrimination.
[0012]
FIG. 1 is a plan view showing a schematic configuration of an image reading device of a bill validating device. In FIG. 1, a plurality of rollers 3 a, 3 b, 3 c for transporting banknotes 2 are arranged on the transport path 1. If the banknote 2 is conveyed from the left side of the figure, an inrush sensor 4 is disposed on the upstream side of the conveyance path 1 and an imaging sensor 5 is disposed on the downstream side.
[0013]
The intrusion sensor 4 is constituted by, for example, a photocoupler composed of a light emitting element and a light receiving element, and the light from the light emitting element received by the light receiving element after the banknote 2 conveyed between the light emitting element and the light receiving element is inserted. Is interrupted, the entry of the banknote 2 is detected, and the banknote 2 passes through the entry sensor 4 and the light receiving element receives light emitted from the light emitting element again to detect the protrusion of the banknote 2. .
[0014]
The imaging sensor 5 is constituted by, for example, a CCD (charge coupled device) line sensor and, for example, a light source of an LED array in which a plurality of light emitting diodes (LEDs) are arranged in a line. The imaging sensor 5 is arranged so as to scan in a direction perpendicular to the transport direction, and sequentially reads the images of the banknotes 2 by controlling the lighting of the light source while transporting the banknotes 2. Further, the light source and the CCD line sensor are arranged on both sides (upper and lower) so as to sandwich the banknote 2. This CCD line sensor detects light from a light source directly incident without passing through the banknote 2 as a transparent pixel having a large pixel value, and detects the incident light transmitted through the banknote 2 with a small pixel value. It is detected as a non-transparent pixel. Therefore, if all the pixel values on a certain scan line indicate a saturated value, it can be determined that the banknote 2 has not arrived or passed on the CCD line sensor, and even if one pixel indicates the value of a non-translucent pixel, It can be determined that there is a bill 2 on the CCD line sensor.
[0015]
FIG. 2 is a diagram for explaining image reading of banknotes. When the bills pass through the image sensor 5, they are not in a straight state with respect to the transport direction, but are mostly skewed, and the resulting bill image 10 is also skewed. Further, in the case of banknotes in circulation, the square portion is particularly easily damaged, and is often bent or damaged. Therefore, for example, areas 11a, 11b, 11c, and 11d in the range from square to 20 mm of the banknote are necessary for the discrimination process, but are not reliable areas particularly as image data for angle calculation. The calculation is indefinite. Therefore, the image data for calculating the skew angle is acquired from the central region 12 excluding the scan region including the regions 11a, 11b, 11c, and 11d.
[0016]
Here, the pre-discrimination process from the skew correction of the banknote image 10 to the extraction of the banknote image data will be briefly described. When the banknote 2 is conveyed from the upstream and the leading edge of the banknote 2 (upper edge of the banknote image 10 in FIG. 2) reaches the rush sensor 4, the rush sensor 4 detects “rushing” and the leading edge of the banknote 2. “Imaging” is started just before the image sensor 5 approaches the imaging sensor 5. The bill 2 is sequentially scanned from the leading edge while being conveyed, and shading correction is performed simultaneously with the acquisition of pixel data. This shading correction will be described later. When the trailing edge of the banknote 2 comes out of the intrusion sensor 4 and “protrusion” is detected, the longitudinal length of the banknote is calculated here, and the time until the remaining 20 mm angle calculation indefinite region is calculated. Subsequently, when scanning by the imaging sensor 5 is performed and the remaining 20 mm is reached, the width and skew angle of the banknote 2 are calculated for the scan data in the region 12 from the scan data acquired so far. That is, the start coordinates and the end coordinates are obtained by obtaining the start coordinates that first change from the light-transmitting pixels to the non-light-transmitting pixels and the end coordinates that end from the non-light-transmitting pixels to the light-transmitting pixels in the scan direction on each scan line The width of the banknote on each scan line is acquired from the difference between and the mode value of the width. This mode value is a value that can be regarded as representing the width of the banknotes excluding the missing part of the long side of the banknote 2. Next, among the scan lines having the mode value, for example, a combination of several high-order lines and a few low-order lines, the distance between these sets of scan lines, and the start coordinates of the non-transparent pixels, for example, of these scan lines The angle is obtained from the above, and the average of the plural sets of angles is defined as the skew angle. Here, since the skew angle has been obtained, skew correction, that is, affine transformation is performed on the scan data acquired from the top of the banknote 2 to the remaining 20 mm. In parallel with the width / angle calculation, imaging of the remaining 20 mm is continued, and affine transformation is also performed on the remaining 20 mm of scan data when the skew angle is obtained. Since the affine transformation is performed at a higher speed than the imaging processing, when the trailing edge of the banknote 2 passes through the imaging sensor 5 and all the imaging sensors 5 detect the translucent pixels, that is, when one line saturation is detected, the discrimination is performed. Most of the pre-processing has already been completed, and thereafter, the banknote image 10 is cut out and the pre-discrimination processing is completed.
[0017]
3A and 3B are diagrams for explaining shading correction, in which FIG. 3A shows a configuration example of a light source, FIG. 3B shows a change in brightness in the longitudinal direction of the light source, and FIG. 3C shows an image sensor output after shading correction. The characteristics are shown. As a light source for illuminating the bill being conveyed, an LED array 20 as shown in FIG. This LED array 20 has a configuration in which a plurality of LEDs 21a, 21b,. The light emitted from the LED array 20 has the highest brightness on the optical axis of each LED 21a, 21b,..., 21n, and decreases as the distance from the optical axis increases, as shown in FIG. Therefore, each pixel of the CCD line sensor outputs a different value depending on variations in brightness depending on the location. Therefore, using a reference medium that performs uniform transmission in advance, the pixel value corresponding to the brightness difference is obtained as a coefficient for each pixel of the CCD line sensor, and it corresponds to each pixel output of the CCD line sensor. Multiply the coefficient to do. This process is called shading correction. As shown in (C), the pixel values are uniform if the medium is solid. This shading correction is performed immediately after the pixel value is output from the CCD line sensor.
[0018]
FIG. 4 is a time chart showing the flow of the pre-discrimination process. First, at the time t0, when the leading edge of the conveyed banknote is detected by the rush sensor 4, a rush interrupt process is started, which is determined in advance from the distance between the rush sensor 4 and the image sensor and the transport speed. At time t1 after the time, imaging by the imaging sensor 5 is started. Furthermore, pre-discrimination processing is started at time t2 after a predetermined time. At this time, the shading correction is performed on the pixel data already captured. The shading correction is immediately terminated because the calculation load is not so large, and thereafter, the shading correction is performed almost simultaneously with the imaging.
[0019]
When the rush sensor 4 detects the protrusion of the banknote at time t3, the protrusion interrupt process is started. In the protrusion interruption process, first, the longitudinal length of the bill is calculated, and the time required for the scan to proceed from the longitudinal length to the angle calculation indefinite region, that is, the remaining 20 mm of the bill is calculated. At this time, since the banknote is being conveyed, imaging is continued without interruption. At time t4 when the scan reaches the remaining 20 mm position, the bill width and skew angle are calculated. That is, based on the scan data acquired up to the remaining 20 mm position, the width of the banknote is calculated from the distribution of non-transparent pixel values on each scan line, and the mode value of the banknote width is obtained. A set of scan lines as far as possible from a scan line having a value is selected, and the skew angle is calculated from the distance between the scan lines of the set and the start coordinates of the non-translucent pixels on each scan line. Actually, the angle is obtained for each of a plurality of sets of scan lines, and the skew angle is obtained from the average of these angles.
[0020]
As soon as the skew angle is obtained, skew correction is performed on the scan data acquired up to the remaining 20 mm by affine transformation. Although the affine transformation has a large calculation load, it still ends with a margin until the banknote passes through the image sensor 5. During this time, in the remaining 20 mm imaging, affine transformation is performed while performing imaging and shading correction. At time t5, when the imaging sensor 5 detects one line saturation indicating that there is no banknote on the imaging sensor 5 by passing the banknote through the imaging sensor 5, most of the banknote image data is affine transformed at this time. Therefore, the skew correction of the banknote image data is substantially completed simultaneously with the detection of one line saturation. Specifically, if the conveyance speed is 500 mm / second, the maximum longitudinal length of the banknote is 172 mm, and the distance between the intrusion sensor and the imaging sensor is 35.8 mm, 85.6 from the detection of the protrusion to the one-line saturation of the imaging completion. There is a millisecond time. On the other hand, the width / angle calculation from the detection of the protrusion and the skew correction from that point to 16.2 milliseconds, and the remaining 20 mm of the shading correction and the skew correction need only about 21.8 milliseconds. Processing can be sufficiently performed from detection to completion of imaging. Thereafter, the banknote image data is cut out and the process is shifted to the discrimination process, whereby the discrimination pre-processing is completed.
[0021]
【The invention's effect】
As described above, in the present invention, after detecting the protrusion of the banknote, after imaging up to the angle calculation indefinite region on the end side of the medium, the width / angle is calculated, and the image data captured up to that time is skewed to correct the width / After the angle calculation, the remaining angle calculation indefinite region is configured to perform skew correction while imaging. As a result, all the image data has been substantially corrected for skewing at the time of imaging to the end of the medium, and the time required for the above pre-processing, which has been conventionally required after imaging, is almost eliminated. And can allocate sufficient time for further processing.
[Brief description of the drawings]
FIG. 1 is a plan view showing a schematic configuration of an image reading device of a bill validating device.
FIG. 2 is a diagram for explaining image reading of banknotes.
FIGS. 3A and 3B are diagrams for explaining shading correction, in which FIG. 3A shows a configuration example of a light source, FIG. 3B shows a change in brightness in the longitudinal direction of the light source, and FIG. 3C shows an image sensor after shading correction; Output characteristics are shown.
FIG. 4 is a time chart showing a flow of operation of pre-discrimination processing.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Conveyance path 2 Banknote 3a, 3b, 3c Roller 4 Inrush sensor 5 Imaging sensor

Claims (4)

In an image processing method for a conveyance medium that corrects an inclination state of an image obtained by imaging while conveying a rectangular medium,
After detecting the transport direction tip of the rectangular medium that has been transported by the medium detection sensor, start imaging for each scan line of the rectangular medium after a predetermined time determined by the transport speed and the distance to the image sensor,
After the medium detection sensor detects the trailing end of the rectangular medium in the transport direction, calculates an indeterminate area arrival time until imaging reaches an angle calculation indeterminate area in front of the rectangular medium in the transport direction trailing edge;
After the indefinite region arrival time has elapsed, while performing angle correction of the tilt image with respect to the image data captured up to that point, the remaining angle calculation indefinite region is imaged and the tilt image is angle corrected.
An image processing method for a carrier medium. The image processing method for a transport medium according to claim 1, wherein shading correction is performed on each detection pixel when the image sensor captures each scan line of the rectangular medium. The angle correction of the tilted image up to the angle calculation indefinite region is performed by acquiring a start coordinate that first changes from the background pixel to the medium pixel in the scan direction and a last coordinate that finally changes from the medium pixel to the background pixel on the scan line. Obtain the width of the medium on each scan line from the difference between the start coordinates and the end coordinates, obtain the mode value of the width of the medium of each scan line obtained until reaching the angle calculation indefinite region, For the scan line having the mode value of the width, an inclination angle is calculated from at least one pair of scan lines located at a distance from each other and the start coordinate or the end coordinate, and the angle calculation indefinite region The image processing method for a transport medium according to claim 1, wherein the image obtained so far is affine transformed at the tilt angle. 2. The image of a transport medium according to claim 1, wherein the angle calculation indefinite region is a range in which the rectangular medium is allowed to be folded or missing in the medium discrimination process after the angle correction process of all images. Processing method.

Images