JP4187043B2 - Image processing device - Google Patents

Description

  BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for preventing misuse image processing of a specially printed original in order to prevent faithful copying of an original for image processing, particularly color originals such as banknotes and securities.

  In recent years, with the improvement of the performance of color copiers, the need to develop copiers that have the function of preventing the copying of banknotes, stock certificates, securities such as bonds, commuter passes, admission tickets for events, and vouchers has increased. Yes. The technology related to the copy prevention function of the conventional copying machine can be broadly divided into a technology for providing a function for preventing a normal copy on the original side and a copy-prohibited original by an image reading / detecting device, and a normal copy operation is not performed. It can be divided into technologies that add functions.

As a function for preventing normal copying on the former document side, for example, metal powder is mixed in the document, and a copy image based on metal powder other than the document image is formed by reflection of the metal image of the document image surface irradiation lamp. Selenium photoconductor Techniques such as one that uses blue ink that is difficult to reproduce in a copier using the printer are known.
In addition, the technology for detecting a copy-prohibited original by the latter image reading / detecting device and providing a function for preventing the normal copying operation compares the read data of the original with the basic data in the memory, and can be copied. There are known methods for determining whether or not to perform copying, prohibiting copying, unfixing the output paper, changing the output state of the paper from normal.

Any of these conventional document abuse prevention techniques has advantages and disadvantages, and a normal copy prevention function on the side of the image reading / detecting device is a memory having a huge capacity for storing basic data of a specific document and a read document image. Therefore, there is a problem that the cost becomes high. A technique for solving such problems is proposed in Patent Document 1. That is, a circular image composed of a plurality of concentric codes specifying at least one of the thickness of each circle, the interval between the lines of each circle, or the diameter of each circle is printed on the document surface, and the concentric codes are By recognizing with an image processing apparatus, misuse image processing is prevented.
JP-A-4-296323

By the way, the conventional circular image for preventing the abused image processing is focused on the point that the document can be recognized at any angle. When detecting a circle, only the image data on the main scanning line is referred to. However, image data input from a manuscript contains noise, and in the prior art, this noise has an adverse effect and lowers the accuracy of circle detection.
An object of the present invention is to provide an image processing apparatus capable of performing accurate detection even when noise is included in image data.

In order to achieve the above object, the present invention described in claim 1 is an image processing apparatus that performs image processing on a document image read in units of pixels and outputs the original image. The position calculation means for calculating the distance, the distance calculation means for calculating the distance from the pixel to the midpoint position, and the distance between the pixel at the midpoint position and the pixel at the midpoint position one line before in the sub-scanning direction is accumulated. A means for detecting a length of a straight line connecting the midpoint positions, whether the length of the straight line is equal to a diameter of a circle, and determining a result of the discrimination and a distance from the pixel to the midpoint position. And a circle extraction means for performing circle extraction.
According to the above configuration, the midpoint position between the pixels in the main scanning direction, the distance from the pixel to the midpoint position, the distance between the pixel at the midpoint position in the sub-scanning direction and the pixel at the midpoint position one line before is calculated. A circle image can be reliably detected with high accuracy by obtaining the length of a straight line that accumulates and connects the midpoint positions and uses it as information for circle extraction. Since real-time processing is possible, high-speed processing is possible.

According to the present invention, not only the image data on the main scan line but also the midpoint position obtained by accumulating the distance between the midpoint position pixel calculated in the sub-scan direction and the midpoint position pixel one line before is connected. Since the length of the straight line is also referred to, it is possible to accurately detect the circle even if the image data includes noise.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 shows the configuration of the entire system of the image processing apparatus. The target document is converted into a plurality of image signals by the sensor 1. Each image signal is processed by the specific image processing device 2 in accordance with each image component and input to the threshold processing device 3. The threshold processing device 3 performs threshold processing based on one or a plurality of thresholds. The pixel extraction processing device 4 determines ON / OFF of the pixel from the result of the threshold processing device 3. The midpoint extraction processing device 5 obtains a midpoint between two black pixels in the main scanning direction, detects connectivity of the midpoint position in the sub-scanning direction, and determines the distance from the pixel to the midpoint position. calculate. The circle extraction processing device 6 identifies a specific circle while accumulating the results of the midpoint extraction processing device 5 in the sub-scanning direction.

  Next, the above processes will be described in detail. FIG. 2 shows the configuration of the input process. The document is read by the sensor 1 and processed as a plurality of image components. Here, the case of RGB three components will be described. In general, since an input image has a blur or the like coming from an optical system, gradation correction / image enhancement is performed. Since the purpose of this process is to extract characteristic pixels rather than image reproduction, image processing specialized for a specific color is effective. For example, in the specific image processing apparatus 2, the R component performs blurring, the G component performs image enhancement, the B component performs through processing, and the like.

The data processed by the specific image processing device 2 is subjected to threshold processing by the threshold processing device 3, respectively. The threshold processing device 3 performs processing with one or a plurality of thresholds. In general threshold processing, THr <Dr THg <Dg THb <Db according to threshold values THr, THg, THb.
In each case, the output signal is turned ON. In addition, by defining the color components, two threshold values are provided, and THr1 <Dr <THr2 THg1 <Dg <THg2
THb1 <Db <THb2
In this case, the output signal can be turned ON.

Furthermore, THr1 <Dr <THr2 THr3 <Dr <THr4 using a plurality of threshold values
THg1 <Dg <THg2 THg3 <Dg <THg4
THb1 <Db <THb2 THb3 <Db <THb4
It is also possible to perform threshold processing with steps as described above.

  In this embodiment, the output from the sensor has been described as R, G, B. However, the signal itself can be realized with the same configuration, whether it is a single signal (for example, only the G component) or a plurality of signals. Further, the signal is not limited to R, G, and B, and the same effect can be obtained with signals such as luminance signals L, a, and b.

The signal subjected to the threshold processing is sent to the pixel extraction processing device 4, and pixel extraction is performed based on the data from the threshold processing device 3 of each signal component. The pixel extraction processing device 4 is generally output as a signal out = Ro & Go & Bo as a result of threshold processing, but by defining the color, out = Ro | Go | Bo out = Ro | (Go & Bo)
out = Ro & Go & Bo
Only a specific color can be extracted by the operation as described above. In addition, the accuracy of data can be given by performing arithmetic operations instead of logical operations.

  FIG. 2 shows a case where a single threshold processing device is provided for each image signal. However, a plurality of threshold processing devices and pixel extraction processing devices may be provided as shown in FIG. Conceivable. In the configuration shown in FIG. 3, pixel extraction can be performed for each color, so that accurate extraction can be performed even when the colors are mottled.

  The concept of circle extraction will be described with reference to FIG. The feature of the circle in the main scanning direction does not change no matter how the document is placed. Here, it is assumed that an inner frame (black pixel) of a circle is extracted while scanning in the main scanning direction, and a midpoint between black pixels is obtained. When this is processed continuously in the sub-scanning direction, the midpoint of the circle is a straight line with respect to the sub-scanning direction, and its length is equal to the maximum length (diameter) in the main scanning direction. A circle is extracted using these features.

Next, a middle point extraction method and a circle extraction method mainly for detecting the continuity of the middle point in the sub-scanning direction will be described. FIG. 5 is an explanatory diagram of the midpoint detection method. The illustrated one is a general example, and when there are four black pixel blocks in the main scanning direction, L1, L2, and L3 are measured, respectively, and midpoint information is also measured at the same time.
A basic technique for midpoint detection will be described with reference to FIG. A point from a black pixel to a white pixel is detected as an inner frame of the circle (S1), and is set as a starting point. The counter is turned ON together with the detection of the start point (S2). Then, a pixel that is a black pixel from a white pixel that is an end point is detected (S3), and the pixels up to the pixel are counted (S4). After the counter is output, the counter is reset (S5). Since the midpoint is half of the count value, the position of the midpoint can be obtained immediately by adding a binary counter. This configuration is an effective method when there is no noise in the target image.

On the other hand, a case where noise is taken into account will be described. FIG. 7 shows a case where noise exists in the circle. The length across the circle is actually L3, but because there is noise, it is divided into L1 and L2, and the exact midpoint of the circle cannot be extracted. However, if the midpoints between all the black pixels in the main scanning direction are obtained in order to deal with noise, real-time processing becomes difficult due to the increase of the midpoints as shown in FIG. In addition, the hardware cost for the midpoint processing also increases, which affects the feasibility / performance and the like in the circle extraction.
Although it is difficult to solve the problem of accurately extracting the midpoint when there is noise in this way by processing only in the main scanning direction, it is possible to cope with it by using the characteristics of the circle in the sub-scanning direction together. .

Next, the midpoint processing method will be described. FIG. 9 shows the state of the midpoints of the distances th1 to th5 when scanning in the raster direction with respect to a circle, a triangle, and a quadrangle. In this case, everything becomes a straight line, and misidentification occurs when this is identified as a circle. Here, before extracting the midpoint, the distance information of the midpoint of the previous line is used. In the case of the circle, the first of distance in accordance with the advance in the sub-scanning direction, which is near 0, last ing and near zero gradually turn to gradually boundary of the increase to go a certain value (diameter) decreases It shows such characteristics. On the other hand, in the case of a triangle, it only increases gradually, and the rectangle only shows a constant value. Since triangles and quadrangles are not always input at such positions, they actually show a more complicated state. On the other hand, since the circle shows the same characteristic regardless of the positional relationship, the midpoint is obtained using this characteristic. Therefore, it is possible to perform processing while limiting target points in a circle. A block diagram in that case is shown in FIG. The start point, end point, and midpoint are detected in the same manner as in FIG.

  The circle is scanned to extract pixels (S10), the start point is detected (S11), and the start point is stored (S12). Similarly, the end point is detected (S13) and stored (S14). A midpoint is detected from the stored start and end points (S15), and the distance between this midpoint pixel and the neighboring midpoint pixel one line before is compared (S16). The feature of the circle is judged from the comparison result (S17). If there is a circle feature, the detected midpoint is stored (S18), and if there is no circle feature, the detected midpoint is stored as invalid ( S19). Then, the process returns to step 11 to process the data by the next scan.

The processing so far has been performed on the entire circle, but only a partial region of the circle may be used as shown in FIG. Further, as shown in FIG. 11B, the sub-scanning direction may be skipped.
As described above, the midpoint detection method using the information in the sub-scanning direction has been described. However, when the above processing is performed, the midpoint is a straight line as shown in FIG. In addition to this condition, it can be determined whether or not it is a circle by determining whether or not the length of the straight line is equal to the diameter of the circle.

  Here, the case where it implement | achieves with the memory for 1 line is demonstrated as one Example of the circle extraction method. FIG. 13 shows the concept of processing. FIG. 14 shows the flow. Note that the method described so far is used as the midpoint extraction method, and here, attention is paid to the pixel detected as the midpoint. When the target pixel is black and the same pixel (neighboring) in the previous line is also a black pixel, addition (+1) is performed. With this processing, the length of the straight line in the vertical direction is stored in the memory. When the pixel of interest is white and the same pixel (neighboring) in the previous line is a black pixel, the memory value so far is examined, and when it is close to the length of the diameter stored in advance, it is determined as a circle candidate. If not, clear the memory value. Circle candidates can be determined by the above processing. In the illustrated example, the number of pixels at the midpoint is “46”.

In FIG. 14, when a middle point pixel is detected in a certain line scan (S20), when the detected pixel is a white pixel, it is determined whether the same pixel in the previous line is black (S21), If so, return to step 20 to detect the next pixel. On the other hand, if the same pixel in the previous line is black, it is determined whether the midpoint length is equal to the threshold value (S22). When the midpoint length is equal to the threshold value, a circle candidate signal is output as a circle candidate (S23), and the next pixel detection is performed.
Returning to the beginning, when a black pixel is detected in step 20, it is determined whether or not the same pixel in the previous line is black (S24). From this discrimination result, the number of pixels is added when the same pixel in the previous line is black (S25), and when the same pixel is white, the number of pixels is set to "1" (S26) and the process returns to step 20.

Application Example 1: When there are a plurality of circles as identification targets As shown in FIG. 15, an extraction method when there are a plurality of circles in a document will be described. Since a plurality of pieces of information are used to determine whether the document is a copy-prohibited document, reliability is improved. It is also possible to provide information by obtaining the arrangement relation and size of a plurality of circles. In FIG. 16, the means (circle detection processing device) described in the above embodiment is used to identify a circle. The circle detection processing device 10 detects circle information, such as the center position and diameter, and sends the circle information to the multiple circle detection processing device 11. The multi-circle detection processing device comprehensively determines the arrangement relationship / size / number of circles based on the circle information, and outputs the result when it is determined that the arrangement relationship is prohibited from copying. To do.

Application example 2: When a specific color is given to a part of a line constituting a circle As shown in FIG. 17, adding a specific color to a circle increases information, thereby improving the reliability of circle extraction. . There is also an advantage that the inclination of the document can be detected at the same time. This embodiment uses the basic method of the circle extraction processing apparatus. In FIG. 18, the image signal output from the sensor 1 is sent to the specific color detection processing device 7, compared with the color stored in advance, and if it matches the specific color, the signal is sent to the circle extraction processing device 6. . When the pixel determined to be the specific color matches one of the pixels at both ends for obtaining the midpoint of the circle, the position where the specific color appears is stored. Since a specific color is added to the circle to be extracted in advance, this circle is a circle to be extracted only when it is determined as a circle based on the midpoint information and further determined that a specific color has been added. Is determined.
If the circle extraction processing device 6 determines that it is a circle, it outputs the position of the center of the circle and specific color information. The specific color information may be a combination of color, absolute position, relative position, angle from the center, and the like.
Application Example 3: When a specific color is applied to a line forming a circle As shown in FIG. 19, a specific color is arranged on a line forming a circle. The color scheme is meaningful when it is decomposed into histograms of the respective colors, and there is a degree of freedom in the color scheme as long as the rules are observed. Because information increases, the reliability of circle extraction is improved.

This embodiment is used as a basic method of circle extraction. However, in order to calculate the histogram, in FIG. 20, a buffer memory (line memory) 12 having a diameter of about a circle in the sub-scanning direction is required. When a circle candidate is extracted by the circle detection processing device 10, the histogram calculation processing device 13 calculates a histogram of the region using the binary image of the circle as a mask image. When the distribution state of the histogram matches the pattern stored in advance sent to the determination processing device 14, it is determined that the circle is the circle of the specific image.
When a circle is detected, the histogram of each color of the circle is examined, and it is determined that it is a predetermined circle only when it is the same as that stored in advance. Since there are two stages of identification, misidentification can be reduced. Further, by storing a plurality of histogram information, other information can be embedded. In this description, the histogram in the RGB space is obtained, but other color spaces may be used.

It is a figure which shows the whole structure. It is a figure which shows the detailed structure of an input process. It is a figure which shows the structure of a pixel extraction process. It is a figure which shows the concept of a circle detection. It is a figure explaining the detection method of a midpoint. It is a block diagram of a midpoint detection. It is a figure explaining the midpoint detection in case noise exists. It is a figure which shows the result of the midpoint detection when noise exists. It is a figure which shows the characteristic of the midpoint of a circle. It is a figure which shows the structure of the midpoint detection method using the information of a subscanning direction. It is a figure explaining the other method of a midpoint detection. It is a figure explaining the concept of circle detection. It is a figure explaining circle detection. It is a figure which shows the process of a circle detection. It is a figure explaining the case where there are a plurality of circles. It is a figure which shows the structure of a circle detection in case there exist a some circle. It is a figure explaining the case where a specific color is attached to a part of a circle. It is a figure which shows the structure of the circle extraction when a specific color is attached to a part of the circle. It is a figure explaining the case where a circle is colored with a specific histogram. It is a figure which shows the structure of the circle | round | yen detection at the time of coloring with a specific histogram to a circle | round | yen.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Sensor, 2 ... Specific image processing apparatus, 3 ... Threshold processing apparatus, 4 ... Pixel extraction processing apparatus, 5 ... Midpoint extraction processing apparatus, 6 ... Circle extraction processing apparatus, 7 ... Specific color detection processing apparatus, 10 ... Circle detection processing device, 11 ... Multiple circle detection processing device, 12 ... Line memory, 13 ... Histogram calculation processing device, 14 ... Judgment processing device

Claims (1)

In an image processing apparatus that performs image processing and outputs a document image read in pixel units,
Position calculating means for calculating the midpoint position of the pixels in the main scanning direction;
Distance calculating means for calculating the distance from the pixel to the midpoint position;
Means for accumulating a distance between a pixel at the midpoint position in the sub-scanning direction and a pixel at the midpoint position one line before, and detecting a length of a straight line connecting the midpoint positions ;
A circle extracting means for determining whether or not the length of the straight line is equal to the diameter of the circle, and performing a circle extraction based on the determination result and the distance from the pixel to the midpoint position;
An image processing apparatus comprising: