JP4321666B2 - Image data correction device - Google Patents

Description

  The present invention relates to an image data correction apparatus for correcting the brightness of image data obtained by reading a form.

  Conventionally, in an OCR (character reading device) or the like, a form to be read is irradiated with light, the intensity of the reflected light is read by a CCD (charge coupled device) sensor or the like, and the level of the read light is binary with a predetermined threshold value. To perform processing such as character recognition.

JP-A-8-251402

However, the conventional OCR and the like have the following problems.
For example, when reading a form using a CCD sensor such as a digital camera, the intensity of light reflected from the form changes due to the influence of extraneous light, etc., and the level of light detected by this CCD sensor shifts as a whole, In some cases, binarization could not be performed correctly. In some cases, a shadow may enter a part of the form, resulting in partial reading errors.

  The present invention solves the problems of the prior art and provides an image data correction apparatus that corrects the luminance distribution of image data obtained by reading a form to a predetermined level.

  In order to solve the above-described problems, the present invention provides an image data correction apparatus in which an image form to be read is decomposed into pixels and optically read, and the luminance k of each pixel coordinate (x, y) on the form is read. Image reading means for storing (x, y) as image data in an image memory, and dividing the image data into a plurality of rectangular image blocks each consisting of m vertical pixels and n horizontal pixels, and the maximum number of pixels in each image block Block image generation means for generating block image data having the luminance P (X, Y) of the central coordinates (X, Y) of the image block, and the luminance k (x, y) of the pixel stored in the image memory ) Is the luminance P1 (Xa, Ya), P2 (Xa, Yb), P3 (Xb, Ya), P4 (Xb, Yb) of the central coordinates of the four image blocks surrounding the pixel in the block image data. ) And a predetermined reference brightness M are used. And K (x, y) = k (x, y) + M-{(Xb-x) (Yb-y) P1 + (Xb-x) (y-Ya) P2 + (x-Xa) (Yb -Y) P3 + (x-Xa) (y-Ya) P4} / {(Xb-Xa) (Yb-Ya)} is corrected in accordance with luminance correction means for calculating corrected luminance K (x, y). ing.

According to the present invention, the following operation is performed.
The image data of the form read by being decomposed into pixels by the image reading means is divided into m × n pixel rectangular image blocks by the block image generating means, and the maximum luminance of the pixels in each image block is determined by the image block. It is regarded as the brightness of the center coordinate. Further, in the luminance correction means, the luminance of each pixel of the image data is corrected based on a predetermined calculation formula according to the luminance of the center coordinates of the four image blocks surrounding the pixel.

  According to the present invention, the block image generation means for dividing the image data of the form into rectangular image blocks including a plurality of pixels and generating block image data having the maximum luminance as the luminance of the central coordinates, and the block Luminance correction means for correcting the image data by the calculation method based on the distance ratio using the image data is provided.

  In addition, a value detected from a block histogram can be used as a predetermined reference luminance M. In this case, when the shadow portion becomes large, the form background color detected from the histogram becomes a shadow color, and the form background color portion is uniformly corrected to a dark color. Therefore, by correcting the image processed in the second invention with the first invention, it is possible to correct the form background color at a predetermined level even when the shadow range is widened. Therefore, the luminance distribution of the read image data can be corrected to a predetermined level regardless of the reading state of the image reading means.

  The best mode for carrying out the invention will become apparent from the following description of the preferred embodiments when read in conjunction with the accompanying drawings. However, the drawings are only for explanation and do not limit the scope of the present invention.

(Reference example)
FIG. 1 is a configuration diagram of an image data correction apparatus showing a reference example in Embodiment 1 of the present invention.

  This image data correction apparatus has an image reading means (for example, a CCD sensor) 11 that optically separates an image of a form 1 to be read into pixels and reads it optically. The CCD sensor 11 converts the luminance of each pixel constituting the image of the form 1 into, for example, image data of 256 gradations and outputs it. An image memory 12 for storing image data is connected to the output side of the CCD sensor 11.

  A frequency distribution generation means (for example, a frequency distribution generation unit) 13 is connected to the image memory 12. The frequency distribution generation unit 13 generates a luminance frequency distribution of the image data stored in the image memory 12, and a frequency distribution memory for storing the frequency distribution generated on the output side of the frequency distribution generation unit 13. 14 is connected. The frequency distribution memory 14 is connected to a form density detecting means (for example, a form density detecting unit) 15. The form density detection unit 15 detects the background color density of the form 1 based on the frequency distribution stored in the frequency distribution memory 14. The form density detecting unit 15 detects a gradation close to the white level and having the highest frequency as the background color density of the form 1.

  The form density detection unit 15 is connected to a luminance correction unit (for example, a luminance correction unit) 16, and the luminance correction unit 16 is further connected to an image memory 12 and a correction value table 17. The brightness correction unit 16 obtains a correction value for each pixel based on the ratio between the brightness of the pixel stored in the image memory 12 and the background color density of the form detected by the form density detection unit 15, and the correction value and The luminance of the image data is corrected by multiplying a predetermined reference luminance.

  That is, the luminance correction unit 16 calculates the normalized luminance by calculating the ratio between the luminance of each pixel stored in the image memory 12 and the background color density of the form detected by the form density detection unit 15. Further, the brightness correction unit 16 obtains a correction value for each pixel by referring to the correction value table 17 using the normalized brightness as a variable, and corrects the brightness of the image data by multiplying the correction value by a predetermined reference brightness. To do. Then, the corrected image data DATA is output from the luminance correction unit 16.

  2A and 2B are explanatory diagrams showing an example of the correction value table 17 in FIG. 1, in which FIG. 2A is a linear correction table, and FIG. 2B is positive / negative. A correction table when gamma correction is combined is shown.

  In the linear correction of FIG. 2A, the same correction value is output for the normalized luminance shown on the horizontal axis. In the gamma correction of FIG. 2B, a correction value calculated by a combination of positive and negative gamma functions is output for the normalized luminance shown on the horizontal axis.

  3A and 3B are explanatory diagrams of the operation of FIG. The operation of FIG. 1 will be described below with reference to FIGS. 3 (a) and 3 (b).

  The image of the form 1 to be read is separated into pixels by the CCD sensor 11 and read. The read luminance for each pixel is converted into image data of 256 gradations from the black level (0) to the white level (255) and sequentially stored in the image memory 12. When reading of the form 1 by the CCD sensor 11 is completed, the frequency distribution generation unit 13 is activated.

  The frequency distribution generation unit 13 sequentially reads the luminance of the image data stored in the image memory 12, generates a frequency distribution of pixels for each luminance, and stores it in the frequency distribution memory 14. As shown in FIG. 3A, the frequency distribution of the pixels of the form 1 has a peak of almost black level corresponding to the brightness of characters and the like, and a peak relatively close to white level corresponding to the density of the ground color. ing. When the frequency distribution generation unit 13 finishes generating the frequency distribution of the form 1, the form density detection unit 15 is activated.

  In the form density detection unit 15, the level P (for example, P = 200) that is close to the white level and has the highest pixel frequency is the background color of the form 1 based on the frequency distribution stored in the frequency distribution memory 14. Detected as concentration. When the background density P of the form 1 is detected by the form density detection unit 15, the brightness correction unit 16 is activated.

  The luminance correction unit 16 sequentially reads the luminance k of each pixel stored in the image memory 12 and performs the following correction processes (1) to (3) according to the value of the luminance k. The corrected luminance K is output.

(1) When k ≦ P First, the ratio (= k / P) between the luminance k of the pixel and the density P of the background color of the form is calculated as the normalized luminance. Further, the correction value table 17 is referred to using the normalized luminance as a variable, and a correction value TBL (k / P) for each pixel is obtained according to a predetermined correction method (for example, linear correction, gamma correction, etc.). Further, the correction value TBL (k / P) is multiplied by a predetermined reference luminance M (for example, M = 230). Then, it is output as image data DATA in which the luminance K of the multiplication result is corrected. That is, the corrected luminance K can be expressed as follows.
K = TBL (k / P) × M

(2) When P <k ≦ 255−M + P The luminance k of the pixel is shifted by (M−P). That is, the corrected luminance K can be expressed as follows.
K = k + MP

(3) When k> 255−M + P In all cases, K = 255.

By the correction processing of the luminance correction unit 16, the frequency distribution of the luminance K of the corrected pixel is corrected so that the background color density P matches the reference luminance M, as shown in FIG. Then, the corrected luminance K is output from the luminance correction unit 16 as image data DATA.
As described above, the image data correction apparatus according to the reference example includes the form density detection unit 15 that detects the density P of the ground color of the form 1 based on the read frequency distribution of the luminance of the pixels of the form 1, A luminance correction unit 16 that corrects the luminance of each pixel using the correction value table 17 is provided so that the color density P is converted into the reference luminance M. Accordingly, there is an advantage that the luminance distribution of the image data can be corrected to a predetermined level regardless of the background color of the form 1 or the reading state of the CCD sensor 11.

(Configuration of Example 1)
FIG. 4 is a block diagram of an image data correction apparatus showing an embodiment of the present invention. Elements common to those in FIG. 1 are denoted by common reference numerals.

  This image data correction apparatus has a CCD sensor 11 that reads an image of a form 1 to be read by dividing it into, for example, pixels of 3000 × 2000 pixels. The CCD sensor 11 outputs the luminance of each pixel constituting the image of the form 1 as, for example, 256-gradation image data. An image memory 12 for storing image data is connected to the output side of the CCD sensor 11.

  A block image generating means (for example, an image dividing unit) 18 is connected to the image memory 12. The image dividing unit 18 divides the image data stored in the image memory 12 into 30 × 20 rectangular image blocks having m (for example, 100) pixels and n (for example, 100) pixels. is there. Further, the image dividing unit 18 generates block image data in which the maximum luminance of the pixels in each image block obtained by the division is the luminance P (X, Y) of the center coordinates (X, Y) of the image block. It is supposed to be. A block memory 19 for storing the generated block image data is connected to the output side of the image dividing unit 18.

  The block memory 19 is connected with a luminance correction means (for example, a luminance correction unit) 20, and the luminance memory 20 is further connected with an image memory 12. The luminance correction unit 20 determines the luminance k (x, y) of the pixel at each pixel coordinate (x, y) stored in the image memory 12 as the center of the four image blocks surrounding the pixel in the block image data. Using the luminances P1, P2, P3, and P4 of the coordinates, correction is performed by a calculation method based on a distance ratio.

In other words, the luminance correction unit 20 uses the predetermined reference luminance M representing the correction position (that is, luminance) on the histogram of the form background color as the center coordinates of the four pixel blocks P1 (Xa, Ya), P2 ( Xa, Yb), P3 (Xb, Ya), and P4 (Xb, Yb) are used to calculate the corrected luminance K (x, y) according to the following equation (1).
K (x, y) = k (x, y) + M-{(Xb-x) (Yb-y) P1 + (Xb-x) (y-Ya) P2 + (x-Xa) (Yb-y) P3 + ( x-Xa) (y-Ya) P4} / {(Xb-Xa) (Yb-Ya)} (1)

  Then, the corrected brightness K (x, y) image data DATA is output from the brightness correction unit 20.

(Operation of Example 1)
5A to 5C are explanatory diagrams of the operation of FIG. The operation of FIG. 4 will be described below with reference to FIGS. 5 (a) to 5 (c).

  The image of the form 1 to be read is separated into pixels by the CCD sensor 11 and read. The read luminance for each pixel is converted into image data of 256 gradations from the black level (0) to the white level (255) and sequentially stored in the image memory 12. When the reading of the form 1 by the CCD sensor 11 is completed, the image dividing unit 18 is activated.

  In the image dividing unit 18, the image data stored in the image memory 12 is divided into a grid pattern of vertical and horizontal lines every 100 pixels in the vertical and horizontal directions, and is divided into 30 × 20 rectangular image blocks. FIG. 5A schematically shows the image data divided by the image dividing unit 18 in a simplified manner. FIG. 5A shows image data in which the central portion is darkened by the shadow (the luminance is lowered).

  Further, the image dividing unit 18 generates block image data with the maximum luminance of the pixels in each image block formed by the division as the luminance P (X, Y) of the center coordinates (X, Y) of the image block. The The generated block image data is stored in the block memory 19. FIG. 5B schematically shows block image data stored in the block memory 19. In FIG. 5B, since the luminance of the image block in the central portion is low due to the influence of the shadow, the luminance of this image block is lower than that of the surrounding image blocks. When block image data is generated by the image dividing unit 18, the luminance correction unit 20 is activated.

  The brightness correction unit 20 sequentially reads the brightness k (x, y) of each pixel coordinate (x, y) in the image memory 12, and based on the block image data in the block memory 19, this brightness k (x , Y) is corrected. That is, in the brightness correction unit 20, the center coordinates (Xa, Ya), (Xa, Yb), (Xb, Ya), (X) of the image block corresponding to the pixel coordinates (x, y) read from the image memory 12. Xb, Yb) is calculated. As shown in FIG. 5C, the X coordinates Xa and Xb are the X coordinate values of the vertical lines on both sides of the correction target pixel among the vertical lines dividing the image data. Similarly, the Y coordinates Ya and Yb are the Y coordinate values of the upper and lower horizontal lines of the correction target pixel among the horizontal lines dividing the image data.

  Next, the luminance P1 (Xa, Ya), P2 (Xa, Yb), P3 (Xb, Ya), and P4 (Xb, Yb) of each image block at these center coordinates are read from the block memory 19. Then, the luminance k (x, y) of the correction target pixel is corrected by the equation (1), and the luminance K (x, y) is calculated. The calculated brightness K (x, y) is sequentially output as corrected image data DATA.

  When the correction target pixel is a peripheral pixel of the form 1, the center coordinates of the four image blocks corresponding thereto cannot be obtained. In such a case, the processing can be performed on the assumption that an image block having the same luminance exists, but the luminance as it is may be output as the image data DATA without performing the correction processing.

(Effect of Example 1)
As described above, the image data correction apparatus according to the first embodiment divides the read image data of the form 1 into rectangular image blocks including a plurality of pixels, and uses the maximum luminance as the central coordinate luminance. An image dividing unit 18 that generates image data, and a luminance correction unit 20 that corrects the image data by a calculation method based on a distance ratio using the block image data are provided. Thereby, the pixels of the image block without shadow are hardly corrected. On the other hand, the darkened part is affected by the brightness of the surrounding image blocks, and is corrected to a brightness close to the ground color of the form without shadows. Therefore, there is an advantage that the luminance distribution of the read image data can be corrected to a predetermined level regardless of the reading state of the CCD sensor 11.

(Modification)
In addition, this invention is not limited to the said reference example and embodiment, A various deformation | transformation is possible. Examples of such modifications include the following (a) to (e).

  (A) The gradation of the image data is not limited to 256. The present invention can be similarly applied to arbitrary gradations such as 16, 64, 128, and 512.

  (B) The correction in the correction table 17 in FIG. 1 is not limited to linear correction or gamma correction.

  (C) Although the correction table 17 in which the conversion table for correction is stored in advance is used in FIG. 1, the calculation may be performed using a correction function without using the correction table 17. For example, in the case of linear correction, the correction table 17 is not necessary.

  (D) In the description of the operation in FIG. 1, it has been described that the luminance correction unit 16 performs the luminance level correction processing for each pixel stored in the image memory 12. For example, the luminance level stored in the image memory 12 is described. Each correction value for 0 to 255 may be calculated in advance, and the luminance level may be replaced with this correction value. This makes it possible to reduce the processing time in the case of bitmap image processing using a lookup table.

  (E) The size of the image block divided by the image dividing unit 18 in FIG. 2 is not limited to 100 × 100 pixels. An appropriate value can be used according to the actual use environment.

It is a block diagram of the image data correction apparatus which shows the reference example of this invention. It is explanatory drawing which shows the example of the correction value table 17 in FIG. It is explanatory drawing of the operation | movement of FIG. 1 is a configuration diagram of an image data correction apparatus showing an embodiment of the present invention. It is explanatory drawing of the operation | movement of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Form 11 CCD sensor 12 Image memory 13 Frequency distribution production | generation part 14 Frequency distribution memory 15 Form density detection part 16, 20 Brightness correction part 17 Correction value table 18 Image division part 19 Block memory

Claims (1)

Image reading means for optically reading a form to be read into pixels and storing the luminance k (x, y) of each pixel coordinate (x, y) on the form as image data in an image memory; ,
The image data is divided into a plurality of rectangular image blocks each consisting of m vertical pixels and n horizontal pixels, and the maximum luminance of the pixels in each image block is set to the luminance P (X, Y) of the center coordinates (X, Y) of the image block. Y) block image generation means for generating block image data;
The luminance k (x, y) of the pixel stored in the image memory is used as the luminance P1 (Xa, Ya), P2 (Xa, X) of the central coordinates of the four image blocks surrounding the pixel in the block image data. Yb), P3 (Xb, Ya), P4 (Xb, Yb) and a predetermined reference luminance M,
K (x, y) = k (x, y) + M-{(Xb-x) (Yb-y) P1 + (Xb-x)
(Y−Ya) P2 + (x−Xa) (Yb−y) P3 + (x−Xa) (y−Ya) P4} / {(
Xb−Xa) (Yb−Ya)}
Brightness correction means for correcting according to the above and calculating the corrected brightness K (x, y),
An image data correction apparatus comprising:

Images