JPH0779325A - Picture reader - Google Patents

Abstract

PURPOSE:To attain accurate height measurement and obtain a beautiful and focused picture without distortion by detecting and correcting a fault of height distribution data of an original measured for focusing and distortion correction processing. CONSTITUTION:A mirror image 11 of an original 10 is formed by a range finding mirror 5 by placing the original 10 to a predetermined position and a mirror image of an original end face is formed continuously with the original image on an extension line of an image of the original platen 1. Then an image pickup means 7 subscans the original 10 mounted upward to read the original from the upward position and a height measurement means obtains the height distribution of the original in the subscanning direction. A curved point is detected from the obtained height distribution and whether or not the curved point is regular is detected and a point at which the height is rapidly changed is detected from the height distribution. Based on the result of detection, a height point measured in error is detected and the detected height point measured in error is corrected by surrounding correct data. The image pickup means 7 is focused based on the height measurement data after the correction to pick up the original 10 or the picked-up original image is corrected.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image reading apparatus for photographing a book document or the like from above, and measuring the height distribution up to the document surface and performing a correction process of the document image based on the measurement. .

[0002]

2. Description of the Related Art Conventionally, as an image reading apparatus for reading a spatially bent original such as a book from above, for example,
As shown in Japanese Utility Model Application Laid-Open No. 50-78114, there is a camera type in which a line sensor is provided on the film surface instead of the film so as to be capable of scanning movement. In addition, Japanese Patent Laid-Open No. 3-11
As disclosed in Japanese Patent Publication No. 7965, in order to correct the image distortion caused by the bending of the original surface when reading the original, the height of the original surface, that is, the degree of bending of the original is detected, and accordingly, It is known to correct an image by expanding the captured image.

[0003]

However, the above-mentioned Japanese Utility Model Laid-Open No. 50-78114 and Japanese Patent Laid-Open No. 3-117965.
The publication does not disclose focusing at the time of shooting. In addition, erroneous measurement data may be included due to noise, disturbances, etc. when measuring the document surface curvature and height distribution, and even if such erroneous measurement occurs, it is not considered how to deal with it. However, the read image may be erroneously corrected.

The present invention solves the above-mentioned problems, and in an image reading device for reading a spatially curved original such as a book, if there is an abnormality in the height distribution measurement data of the original, the error is measured. It is an object of the present invention to provide an image reading device that can obtain a normal document image by detecting and correcting it.

[0005]

In order to achieve the above object, the present invention is an image reading apparatus for reading an upward document by optical scanning from above, and an image pickup having a line sensor for sub-scanning the document and reading it from above. Means, a height measuring means for obtaining a height distribution of the original in the sub-scanning direction, a first detecting means for detecting an inflection point from the height distribution obtained by the height measuring means, and the inflection point is normal or normal. Of the first to third detecting means, and second detecting means for detecting whether or not the height distribution is abruptly changed by the height measuring means. A fourth detecting means for detecting an erroneous height measuring point based on the detection result and a correcting means for correcting the erroneous height measuring point by the fourth detecting means by correct data of the surroundings are provided, and the corrected height measuring data by the above is obtained. Focusing operation of the image pickup means based on Was taken an original, or is a captured document image that so as to correct.

[0006]

According to the above construction, the image pickup means sub-scans the original placed upward and reads it from above, and the height measuring means obtains the height distribution of the original in the sub-scanning direction. An inflection point is detected from the obtained height distribution, and further, it is detected whether this inflection point is normal or not, and a point where the height changes abruptly from the height distribution is detected. An erroneous measurement height point is detected based on these detection results, and the detected erroneous measurement height point is corrected by correct data of the surroundings. Based on the corrected height measurement data, the image pickup means is brought into focus to photograph the original document or the photographed original image is corrected.

[0007]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows the overall configuration of the image reading apparatus. In the figure, a document such as a book or a file is placed upward on a document table 1 of the image reading apparatus, and an imaging camera unit 2 for reading the document by reading scanning is provided at a position where the document can be read from above. A predetermined space is provided between the document table 1 and the imaging camera unit 2, and a work space is formed by this space. Further, in this apparatus, an illumination unit 3 arranged above the document table 1 for illuminating a document, an operation unit 4 for setting image reading conditions, and the like are provided at the end of the document table 1 at the end of the photographing range. A range-finding mirror 5 that is arranged in the sub-scanning direction and that captures the edge shape of the document is provided, and a control unit (not shown) is provided. The edge shape of the document imaged on the range-finding mirror 5 is read by the image capturing camera unit 2 and converted into the height distance of the document surface from the number of pixels of the image sensor of the image capturing camera unit 2.

FIGS. 2 and 3 respectively show schematic configurations of the apparatus as viewed from the front and the sides. The image pickup camera unit 2 is an optical system for forming a document image on an image pickup element, and a taking lens 6 driven by an autofocus (AF) motor,
CCD line sensor 7 that scans in the sub-scanning direction (direction indicated by an arrow in FIG. 2) on the focal plane on which the document image is formed.
Consists of. A part of the CCD line sensor 7 reads the image of the end shape of the original document reflected on the distance measuring mirror 5. The document 10 placed on the document table 1 is, for example, a book or file in which each page has a columnar shape in which each page is spatially curved when closed at one end and opened left and right. The distance measuring mirror 5 also functions as a stopper for document positioning that extends in the left-right direction on the back side of the document table 1, and is installed at an angle of 45 degrees with respect to the surface of the document table 1. The original 10 is aligned by applying the upper end of the original 10 to the lower end of the distance measuring mirror 5. The document table 1 is colored darker than the background density of the document.

FIG. 4 is a diagram showing the principle of distance measurement.
Is placed at a predetermined position, a mirror image 11 of the original 10 is formed by the distance measuring mirror 5, and a mirror image of the end surface of the original is continuously formed on the extension line of the image of the original table 1 to the original image. It should be noted that the reading range of the image pickup device is shown by a chain line 12. The pixels of the CCD line sensor 7 are indicated by 1 to n.

FIG. 5 shows the image data read in the state of FIG. In the figure, a is a document image, b is a document table image, and c.
Is a portion reflected on the range-finding mirror, d is an image of the end face of the document, and e is a reference for aligning the document. The image d on the end surface of the document is read as if the height of the image was slightly distorted upward due to its height change. Since the document and its end face are illuminated by the illumination unit 3, the entire document is read in white. On the other hand, since the document table 1 is colored, the amount of incident light is small and the image is read in black where the document end face of the distance measuring mirror is not visible.

FIG. 6 shows an output example of one line in the main scanning direction by the CCD line sensor 7. The horizontal axis shows the number of pixels of the line sensor (left ... back side, right ... front side), and the vertical axis shows the sensor surface illuminance. Is the part of the document on which the edge of the document is not visible, is the part of the document that is projected on the ranging mirror,
Is a document portion, and is a document table portion. L1 is the illuminance threshold value, and n1 and n3 are the minimum pixels that pass the threshold value L1.
The maximum value, n2, is the pixel (fixed) corresponding to the document reference position. (N2-n1) is the number of pixels corresponding to the height of the document. Further, (n3-n2-α) is the number of pixels corresponding to the dimension of the document in the front-rear direction. α is a curved portion at the lower end of the document due to a change in the document height, and the document height at this portion (n2-n
It is a value determined by 1). When the CCD line sensor 7 moves in the left-right direction (sub-scanning direction), the value of n1 changes, so that the distribution of distance measurement values in the left-right direction of the document can be obtained. In addition, it is possible to detect the lateral dimension in which the illuminance on the sensor surface is entirely reduced.

FIG. 7 shows a block configuration of a distance measuring circuit in the control unit. The output of the sensor 7 is A / D converted by the A / D converter 21 and then input to the comparator 22. A threshold value is preset in the comparator 22 by the CPU 23. When the output that exceeds the threshold value is input to the comparator 22, the counter 24 is set. The CPU 23 stores the minimum / maximum value of the count value of the counter 24.
From this, the distance corresponding to the document height is calculated and
At the same time, the horizontal dimension of the document can also be obtained.
The CPU 13 uses the sensor moving unit 25 based on the document height distance.
To output a command to the sensor 7 for focusing operation,
In addition, a command is output to control the lighting of the lighting unit 3.

FIG. 8 and FIG. 9 show the results of actually measuring the height distribution of the document in the sub-scanning direction (scanning direction). In addition,
The height is measured during the preliminary scan. Both are height measurement data on the right page of the opened book. Point A in FIG. 8 and point B in FIG. 9 correspond to the binding portion of the book, respectively, and the right side is the right page,
The left side is the left page. Further, the point C in FIG. 8 and the point D in FIG. 9 are the edge portions of the book. FIG. 8 shows a state where the height is normally measured. In FIG. 9, false measurement is made at two points P and Q. The cause of false measurement is electrical noise, distance measuring mirror 5
Dirt, dust, dirt on the top edge of the document, dust,
External light may be incident.

Next, a method of checking height measurement data, that is, a method of detecting and correcting an erroneous height measurement point will be described. The detection and correction processing is performed by the CPU 23 in FIG. 7. The basic idea of detection of erroneous height measurement is (1) the height measurement point changes abruptly compared with the surrounding values, (2) the binding point of a book, the apex of the curve ( The inflection point appears other than the regular inflection point, and the place where either of the above is satisfied is the false high-altitude point.
Both can be detected based on the difference data of the measured values.

FIG. 10 is a schematic flowchart showing a method of detecting and correcting an erroneously measured height point. Hereinafter, this flow will be described. First, difference data is created (# 1). The measured data is TKS (x) [mm] (x = 1,2,…,
N), the difference data is SAB (x) = TKS (x) -TKS (x +
1) (x = 1,2, ..., N-1).

Next, the regular inflection points are extracted (#
2). In # 3 to be described later, the size of the difference data is compared with a certain threshold value, and this threshold value is changed between the normal inflection point and the non-normal inflection point. Therefore, a regular inflection point is extracted.

(1) Check the sign of the adjacent difference data. (x = 1,2, ..., N-2) If the signs of SAB (x) and SAB (x + 1) are different, TKS (x + 1) is regarded as an inflection point. (2) In the case of x = 1, 2, ..., N-4: For the inflection point, the signs of two adjacent difference data are checked. SAB (x + 1)
Is the same sign as both SAB (x + 2) and SAB (x + 3),
TKS (x + 1) is regarded as a regular inflection point (binding part, curved vertex, etc.). If the sign of SAB (X + 1) is different from that of at least one of SAB (x + 2) and SAB (x + 3), TKS (x + 1) is regarded as a non-regular inflection point (inflection point There is a point of high measurement in the vicinity). When x = N-3: Check the sign of one adjacent difference data. If SAB (x + 1) has the same sign as SAB (x + 2), TKS (x + 1) is regarded as a regular inflection point (binding portion, apex of curvature, etc.). If SAB (x + 1) has a different sign from SAB (x + 2), then TKS
(x + 1) is regarded as a non-regular inflection point (there are some points that were falsely measured near the inflection point). When x = N-2: TKS (x + 1) is regarded as a regular inflection point.

As described above, in FIG. 9, point B (the binding portion of the book) and point E (the apex of the curve) are extracted as regular inflection points, and points P, Q, R, S, etc. are not regular inflection points. It is said that

Next, the size of the difference data is checked and corrected (# 3). For the difference data string, the size of the difference data is sequentially checked to check whether there is an erroneous measurement high point. If there is an erroneous measurement high point, correct it each time.

(1) Size check of difference data Absolute value of difference between adjacent difference data | SAB (x + 1) -SAB (x) |
Is compared with some threshold THRE. | SAB (x + 1) -SAB (x) | ≤ THRE: TKS (x + 2) is correct. ｜ SAB (x + 1) -SAB (x) ｜ ＞ If THRE: If TKS (x + 1) is a regular inflection point, TKS (x + 2) is correct (inflection such as binding) There are often large changes in height near the point). TK
When S (x + 1) is not a regular inflection point, TKS (x + 2) is a falsely measured height (it is generally unlikely that the height changes greatly except near the inflection point).

The method of determining the threshold value is as follows. TK
The threshold value is changed depending on whether S (x + 1) is an inflection point (regular or not regular) (it needs to be divided because it is compared with the absolute value of the difference of the difference data). If TKS (x + 1) is not an inflection point: | SAB (x) |> 1 and
If | SAB (x + 1) |> 1, then THRE = SAB (x).
When ｜ SAB (x) ｜ ≦ 1 or ｜ SAB (x + 1) ｜ ≦ 1, TH
RE = 1 (area where height change is small). If TKS (x + 1) is an inflection point: | SAB (x) |> 1 and
When | SAB (x + 1) |> 1, THRE = Max (| SAB (x) |, 2). When | SAB (x) | ≦ 1 or | SAB (x + 1) | ≦ 1, THRE = 2 (area where height change is small). By the above detection method, the point P in FIG. 9 is detected as an erroneous height. Point Q is | SAB (x) -SAB (x + 1) | <THRE and is not detected.

(2) Correction of erroneous height measurement data When the erroneous height measurement is detected, the height measurement value greatly changes. Assuming that there is correct data in the subsequent height measurement data, there is a place where the height measurement value changes in the opposite direction, and in that vicinity there is a height measurement data Tc1 that is close in value to the correct data TcO immediately before the false height measurement data. There should be. The height measurement data Tc1 is regarded as correct data, and the wrong height measurement data between TcO and Tc1 is corrected by linear interpolation. TKS
(x + 3), TKS (x + 4), TKS (x + 5), TKS (x + 6) Search for correct data in this order. If it is not found, it is not possible to correct the incorrect measured height data.

Determination whether TKS (x + 3) is correct data:
If x ≤ N-3 and SAB (x) ≥ 0, and if TKS (x + 3) ≤ TKS (x) and ｜ TKS (x + 3) -TKS (x) ｜ <THRE * 4, then TKS ( x + 3)
Is correct If SAB (x) <0, TKS (x + 3) ≥ TKS (x)
And ｜ TKS (x + 3) -TKS (x) ｜ <THRE * 4, TKS (x
+3) is correct.

Judgment whether TKS (x + 4) is correct data:
When x ≤ N-4, SAB (x) ≥ 0, TKS (x + 4) ≤ TKS (x)
And ｜ TKS (x + 4) -TKS (x) ｜ ＜ THRE * 5, TKS
(x + 4) is correct. If SAB (x) <0, TKS (x + 4) ≥ TKS
(x) and ｜ TKS (x + 4) -TKS (x) ｜ <THRE * 5, T
KS (x + 4) is correct.

Judgment whether TKS (x + 5) is correct data:
If x ≤ N-5, SAB (x) ≥ 0, TKS (x + 5) ≤ TKS (x)
And ｜ TKS (x + 5) -TKS (x) ｜ ＜ THRE * 6, TKS
(x + 5) is correct. If SAB (x) <0, TKS (x + 5) ≥ TKS
(x) and ｜ TKS (x + 5) -TKS (x) ｜ <THRE * 6, T
KS (x + 5) is correct.

Determination whether TKS (x + 6) is correct data:
When x ≤ N-6, SAB (x) ≥ 0, TKS (x + 6) ≤ TKS (x)
And ｜ TKS (x + 6) -TKS (x) ｜ ＜ THRE * 7, TKS
(x + 6) is correct. If SAB (x) <0, TKS (x + 6) ≥ TKS
(x) and ｜ TKS (x + 6) -TKS (x) ｜ ＜ THRE * 7, T
KS (x + 6) is correct. Correction of false measurement height data is performed by linear interpolation. For example, if TKS (x + 4) is correct,

[0027]

[Equation 1] TKS (x + 2) = TKS (x + 1) + {TKS (x + 4) -TKS (x + 1)} / 3 TKS (x + 3) = TKS (x + 1) + { TKS (x + 4) -TKS (x + 1} * 2/3 SAB (x + 1) = TKS (X + 1) -TKS (x + 2) SAB (x + 2) = TKS (X + 2) -TKS (x + 3)

From the above, the T point is judged to be correct data in FIG. Therefore, the corrected P'point data of the P point is TKS (P ') = {TKS (S) + TKS (T)} / 2. Here, TKS (S): the measured value of the S point, and TKS (T): the measured value of the T point.

Next, extraction and correction of non-regular inflection points are performed (# 4). For the data corrected by the size check of the difference data, the regular inflection point is detected again, and the irregular inflection point is corrected. The method of detecting the regular inflection point is the same as described above. The correction method is as follows. If TKS (x + 1) is not a regular inflection point, TKS (x + 2) may also be false measured data, so correct both. The correction is performed by linear interpolation.

[0030]

[Formula 2] TKS (x + 2) = TKS (x + 1) + {TKS (x + 4) -TKS (x + 1} / 3 TKS (x + 3) = TKS (x + 1) + {TKS (x + 4) -TKS (x + 1} * 2/3 SAB (x + 1) = TKS (X + 1) -TKS (x + 2) sAB (x + 2) = TKS (X + 2)- TKS (x + 3)

In FIG. 9, points Q and R are extracted as inflection points which are not normal, and are corrected by the data of points U and V. The corrected Q ', R'point data is TKS (R') = TKS (U) + {TKS (V) -TKS (U)} / 3 TKS (Q ') = TKS (U) + {TKS ( V) -TKS (U)} * 2/3.

Finally, comprehensive evaluation is performed (# 5). Taking lens 6 in real time during the main scan based on the height measurement data
To move the camera so that the document surface is always in focus (AF: Auto Focus). Further, since the photographing magnification also changes as the height of the original changes, the result of the photographing is that the original is distorted. Therefore, the magnification is corrected based on the height measurement data to correct the distortion (bend correction). In AF, the movable range of the taking lens 6 and the taking lens 6
Since there is a magnification correction range in each of the movement followable range and the bending correction, A
F, bend correction is not possible. So A
F, bend correction height limit (Hmm), AF possible gradient limit (S degree) are set, and it is checked whether the height measurement data is within the limit. If it is not within the limit, the CPU 23 displays a warning or makes a sound on the panel of the operation unit 4, and does not perform the main scan.

It is not necessary to check the height measurement data for all the height measurement data, and if the original effective area in the sub-scanning direction is known, only the data within that area may be checked. . In that case, unnecessary calculation is not required, and the calculation speed is increased. A outside the effective area
F. Even if there is height measurement data that exceeds the limit of bend correction, the main scan may be performed without concern. The valid area of the original can be specified by the user using the operation unit 4 or the like, or can be determined by detecting the original size.

The present invention is not limited to the above embodiment, but various modifications can be made. For example, in the above-described embodiment, the distance measurement operation is performed by the prescan before the image pickup operation is performed, but the distance measurement and the focus adjustment may be simultaneously performed during the image pickup operation. In addition to using a single-focus lens to focus only by moving the lens, a zoom lens can be used to correct the magnification in the main scanning direction at the same time as focusing, and the lens and sensor can be moved integrally to focus. It is also conceivable to correct the magnification in the main scanning direction simultaneously with the adjustment.

[0035]

As described above, according to the present invention, in an apparatus for scanning a document such as a book with a line sensor and photographing the document from above, the height of the document measured for focusing and distortion correction processing is high. If there is an abnormality in the distribution data, it is detected and corrected, so there is no need to regard the binding portion of a book or the apex of the curve as an erroneous height measurement point.
Accurate height measurement can be performed and a beautiful image with no distortion can be obtained.

[Brief description of drawings]

FIG. 1 is a perspective view of an image reading apparatus according to an embodiment of the present invention.

FIG. 2 is a schematic configuration diagram of the image reading apparatus as viewed from the front.

FIG. 3 is a schematic configuration diagram of the image reading apparatus viewed from the side.

FIG. 4 is a diagram showing a principle of distance measurement in the image reading apparatus.

FIG. 5 is a diagram showing read image data.

FIG. 6 is a diagram showing an example of output for one line by an image sensor.

FIG. 7 is a block diagram of a distance measuring circuit of the image reading apparatus.

FIG. 8 is a diagram showing height distribution measurement data of a document in the sub-scanning direction.

FIG. 9 is a diagram showing height distribution measurement data of a document in the sub-scanning direction.

FIG. 10 is a flowchart showing a check process of height measurement data.

[Explanation of symbols]

 6 lens 7 CCD line sensor 10 original document 23 CPU

Claims (1)

[Claims] 1. An image reading apparatus for reading an upward document by optical scanning from above, an image pickup unit having a line sensor for sub-scanning the document and reading it from above, and a measuring device for obtaining a height distribution of the document in the sub-scanning direction. High means, first detecting means for detecting an inflection point from the height distribution obtained by the height measuring means, second detecting means for detecting whether the inflection point is normal or not, and the height measuring means Third detecting means for detecting a point where the height changes abruptly from the height distribution obtained in step 4, and fourth detecting means for detecting an erroneously measured height point based on the detection results of the first to third detecting means. And correction means for correcting the false height measurement point by the fourth detection means by correct data of the surroundings, and based on the height measurement data after correction by the above, the image pickup means is brought into focus operation to photograph the original document. Or the captured original image Image reading apparatus is characterized in that as a positive to.