JP2010226689A - Image evaluating device, image evaluating method, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image evaluating device which can accurately detect dust or dirt deposited on a reference white plate or a reflecting mirror in a digital image reader even when noise caused by various sorts of factors is included. <P>SOLUTION: The image evaluating device 1 includes an integrated value calculator 114 for averaging read images of a reference white plate before shading correction for each pixel, an approximation-equation calculator 115 for performing approximation-operation with use of a function for pixel positions, an integrated value corrector 116 for correcting a data arrangement, a numerical value converter 117 for converting the data array to a numeral value, a differential calculator 118 for calculating a difference in the data arrangement, a feature amount calculator 119 for calculating a feature according to the data array, and an evaluated result output section 120 for determining presence or absence of dust or dirt on the reference white plate or a reflecting mirror in a digital image reader on the basis of the feature amount. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image evaluation apparatus, an image evaluation method, and a recording medium for evaluating an image defect obtained from an image reading apparatus such as a scanner or a copying machine.

In a general digital image reading apparatus, light is irradiated onto the surface of an original on which an original image is placed, and the light reflected after hitting the original image is imaged on a solid-state image sensor called a CCD through a reflection mirror and a lens. And has a reading mechanism.
At this time, shading correction and analog gain adjustment are often performed in order to clearly capture the image of the document. However, if dust or dirt adheres to the reference white plate or reflecting mirror used for shading correction or the like, a defective image such as a black stripe or a white stripe is generated at the corresponding position of the read image. In particular, since the CCD of recent image reading apparatuses has been reduced in pixel size, the influence of even minute dust and dirt has increased.
Therefore, ingenuity and measures are taken to reduce defective products by improving the cleanliness of parts and factories mounted on the product of the image reading apparatus.
In order to solve such a problem, in Patent Document 1, it is determined that there is dust or dirt depending on whether or not the raw data of the read image of the reference white board exceeds a certain threshold value. However, recent CCDs have a small pixel size, and the S / N ratio of the image reading signal has become disadvantageous, and the defective part is often buried in noise. In other words, since noise countermeasures are not taken into account in determining whether the obtained waveform data exceeds a certain threshold value, the risk of erroneous detection increases.

  Image data before shading correction is affected by uneven illumination in the illumination system and COS 4 law of the lens system, and the uniformity of the waveform data is not always guaranteed, which also causes false detection. Yes. In addition, a paint chart formed in uniform gray is attached to a dedicated metal plate, and dirt on the reflection mirror and the reference white plate is detected using the read image data after shading correction when this is used as a document. For this detection, labor and cost are required for maintenance and management in order to avoid erroneous detection due to contamination of the chart.

  Therefore, the present invention has been made in view of the above problems, and the problem is that even if noise resulting from various factors is included, dust and dirt adhering to the reference white plate and reflecting mirror of the digital image reading apparatus are included. It is to provide an image evaluation apparatus, an image evaluation method, and a recording medium that can accurately detect the image.

The features of the present invention, which is a means for solving the above problems, are listed below.
An image evaluation apparatus according to the present invention is an image evaluation apparatus for detecting dust or dirt on a reference white plate or a reflection mirror of a digital image reading device, and averages the read image of the reference white plate before shading correction for each pixel. Calculated by an integral value calculation unit that generates a first one-dimensional integral data array, an approximate expression calculation unit that approximates the first one-dimensional integral data array using a function of each pixel position, and the approximate expression calculation unit. An integral value correction unit that corrects the first one-dimensional integral data array using the reciprocal of the second one-dimensional data array, and a numerical value for the third one-dimensional integral data array calculated by the integral value correction unit. A numerical value conversion unit for generating a fourth one-dimensional integral data array by conversion, and calculating a difference between the third one-dimensional integral data array and the fourth one-dimensional integral data array to obtain a fifth one-dimensional data Differences that produce an array A calculation unit, a feature amount calculation unit that calculates a feature amount based on the fifth one-dimensional data array, and a reference white plate or a reflective mirror of the digital image reading device based on the feature amount calculated by the feature amount calculation unit And an evaluation result output unit for determining the presence or absence of dust or dirt on the top.
The image evaluation apparatus according to the present invention is further characterized in that the numerical value conversion unit is a low-pass filter.
The image evaluation method of the present invention is an image evaluation method for detecting dust and dirt on a reference white plate or a reflection mirror of a digital image reading apparatus, and averages the read image of the reference white plate before shading correction for each pixel. An integral value calculating step for generating a first one-dimensional integral data array, an approximate expression calculating step for approximating the first one-dimensional integral data array using a function of each pixel position, and an approximate expression calculating step An integral value correcting step of correcting the first one-dimensional integrated data array using the inverse of the second one-dimensional data array, and a numerical value of the third one-dimensional integrated data array calculated by the integral value correcting step. A numerical value conversion step of converting to generate a fourth one-dimensional integral data array; calculating a difference between the third one-dimensional integral data array and the fourth one-dimensional integral data array; Array Based on the difference calculation step to be generated, the feature amount calculation step for calculating the feature amount based on the fifth one-dimensional data array, and the reference white plate of the digital image reading device based on the feature amount calculated by the feature amount calculation step Or an evaluation result output step for determining the presence or absence of dust or dirt on the reflecting mirror.
The image evaluation method of the present invention is further characterized in that the numerical value conversion step is a low-pass filter.
The recording medium of the present invention is a computer-readable medium on which a program for causing a computer to implement the image evaluation method is recorded.

The present invention, which is a means for solving the above problems, has the following specific effects.
The image evaluation apparatus of the present invention can correct illumination and lens system illuminance unevenness even with image data before shading correction, and can accurately detect dust and dirt on the reference white plate and the reflection mirror.

It is the schematic which shows the structure of the image evaluation apparatus applied to embodiment of this invention. It is a block diagram of the image processing apparatus in an image evaluation apparatus. It is a flowchart showing operation control of an image evaluation apparatus. It is a graph which shows the waveform of the reference | standard white board mounted in the digital image reader. It is a graph which shows the 2nd one-dimensional integral data array Zi produced | generated by the approximate expression calculation part. It is the graph which overwritten the 4th one-dimensional integral data array Bi on the 3rd one-dimensional data array Ai. It is the graph which showed the 5th one-dimensional data array Ci. It is a graph which shows providing a threshold value and distinguishing dirt etc. This is an enlarged view of the circled portion in FIG.

The best mode for carrying out the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic diagram showing a configuration of an image evaluation apparatus according to an embodiment of the present invention.
The image evaluation apparatus 1 includes, for example, an image processing apparatus 11 and an image reading apparatus (scanner) 12 configured to have a predetermined function in a personal computer (PC) having a general configuration. The image evaluation apparatus 1 can further perform an image evaluation method using the image processing apparatus 11 or the like shown in the following block diagram. Further, this image evaluation method may be recorded in a readable manner as a program that can be read by the image processing apparatus 11 on a recording medium (not shown) so as to operate the image evaluation apparatus 1.
FIG. 2 is a block diagram of an image processing apparatus in the image evaluation apparatus.
In the image processing apparatus 11, the image evaluation apparatus 1 is used as a CPU 111 for managing the operation of the entire apparatus, a memory 112 for storing input image data and measurement results as well as a work memory for the CPU 111, and an image An image input unit 113 for reading a reference white board mounted on the reading device 12 as RGB digital image data before shading correction, and averaging the image data for each pixel Xi to obtain a first one-dimensional integrated data array Yi. An integral value calculation unit 114 to be generated, an approximate expression calculation unit 115 that approximates the first one-dimensional integration data array Yi by using a function of each pixel position, and a second value calculated by the approximate expression calculation unit 115 An integration value correction unit 116 that corrects the first one-dimensional integration data array Yi using an inverse of the one-dimensional data array Zi; and the integration value correction unit A numerical conversion unit 117 that numerically converts the third one-dimensional integral data array Ai calculated by 16 to generate a fourth one-dimensional integral data array Bi, and the third one-dimensional integral data array Ai and the fourth A difference calculation unit 118 that calculates a difference of the one-dimensional integral data array Bi of the second and generates a fifth one-dimensional data array Ci, and calculates an image evaluation value from the feature amount of the fifth one-dimensional data array Ci. An image evaluation value calculation unit 119 and an evaluation result output unit 120 for externally outputting the evaluation result to a monitor, a printer, a file, or the like are provided.
Further, the image evaluation apparatus 1 includes a reading device (scanner) 12 for reading an image to be evaluated. The reading device 12 fixes the reading position of a line image sensor composed of a CCD, moves the read document in the sub-scanning direction with respect to the reading position, exposes it by the exposure device, and reflects the reflected light of the document as a document image. The image signal output from the line image sensor, which is read by the line image sensor, is sampled and converted into a digital signal to form read image data.
Furthermore, the image evaluation apparatus 1 can be constructed on a general basis by storing a computer-readable recording medium on which a program to be implemented by a computer is recorded.

FIG. 3 is a flowchart showing operation control of the image evaluation apparatus. Here, the operation control of the image evaluation apparatus 1 is shown from when the reference white plate is read by the reading device 12 until it is determined whether there is dust or dirt.
First, density data is taken in the circumferential direction in which the photoreceptor of the reference white plate to be evaluated rotates (step S1). A reference white board mounted on the digital image reading apparatus is captured by the image input unit 113.
FIG. 4 is a graph showing a waveform of a reference white plate mounted on the digital image reading apparatus.
A reference white plate mounted on the reading device 12 that digitizes and reads an image is captured by the image input unit 113 in the image evaluation device 11. The image data captured by the image input unit 113 is averaged for each pixel by the integration value calculation unit 114 to generate a first one-dimensional integration data array Yi (step S2). The graph shown in FIG. 4 shows the first one-dimensional integral data array Yi obtained by the integral calculator 114 as a waveform. In this waveform, since it is an image before shading correction, the illuminance distribution by the illumination and the lens is not uniform, so that the vertical streak component indicated by the circle ○ is buried in the swell component.
Conventionally, a defect such as dust is detected from this image data or the first one-dimensional integral data array Yi. As shown in FIG. 4, the presence of dust or the like could be found by showing a high image density as indicated by a circle ◯. However, even in the swell component, there is a portion with a high output level. If the output level in the circle ○ is defined as dust, the swell component portion is also all dirty with dust.

ここで、ｎは多項式で近似する場合の次数、ａとｋは、近似式の係数を表す。 Next, in accordance with the flowchart of FIG. 3, a method of accurately extracting vertical stripes by removing the waviness component from image data of defects such as dust will be described.
First, an average value Y of the first one-dimensional integral data array Yi is calculated from the first one-dimensional integral data array Yi (step S3).
Assuming that the first one-dimensional integration data corresponding to each pixel Xi is Yi, the first waveform can be approximated by applying a polynomial approximation using the following equation (1) (step S4).

Here, n represents the order when approximated by a polynomial, and a and k represent coefficients of the approximate expression.

FIG. 5 is a graph Zi showing the second one-dimensional integral data array generated by the approximate expression calculation unit. As shown in FIG. 5, the approximate expression calculation unit 115 performs fourth-order polynomial approximation using a plurality of sets of (Xi, Yi) data in the pixel i to generate a second one-dimensional integral data array Zi. It is a result. In the graph, the second one-dimensional integral data array Zi is overwritten on the first one-dimensional integral data array Yi.
Here, n = 4, but the order is not limited, and it is needless to say that an order capable of removing the swell component may be appropriately selected. In order to perform vertical streak detection, the second one-dimensional integrated data array Zi is an unnecessary shading component when detecting dust and dirt, so a method of removing it from the first one-dimensional integrated data array Yi is used. Think.

Next, the integral correction unit 116 obtains the reciprocal number Zi of the second one-dimensional integral data array Yi (step 5). The second one-dimensional integral data array Zi is multiplied by the first one-dimensional integral data array Yi for each corresponding pixel to calculate a third one-dimensional integral data array Ai (step 6). Next, the average value A of the third one-dimensional integral data array Ai is calculated (step 7). Next, the third one-dimensional integration data array Ai is corrected with the average value Y of the first one-dimensional integration data array and the average value Y of the first one-dimensional integration data array Yi, and the third one-dimensional integration data array Ai is corrected. The data array A′i is calculated (step 8). Note that, simply by multiplying the reciprocal, the average value of the correction data changes, so the average value Y of the first one-dimensional integral data array Yi and the average value A of the correction data A′i are the same value. Thus, level correction is performed by multiplying the correction constant: Y / A.
FIG. 6 is a graph in which the fourth one-dimensional integrated data array Bi is overwritten on the third one-dimensional data array A′i. Here, FIG. 6 shows the third one-dimensional integral data array A′i corrected by the integral value correcting unit 116. In this example, vertical streaks can be extracted from the waveform of FIG. 5 using the constant threshold value in the third one-dimensional integrated data array A′i.

Here, a method of further removing noise is considered.
That is, the numerical value conversion unit 117 performs a median filter on the third one-dimensional integral data array A′i to generate a fourth one-dimensional integral data array Bi (step S9).
The median filter is a process in which output values in a local region are arranged in ascending order, and a median value is output at the center of the region, and has an effect of removing only noise (random noise) while keeping an edge.
In FIG. 6, the fourth one-dimensional integrated data array Bi is overwritten on the third one-dimensional data array A′i.
Next, the difference calculation unit 118 generates a fifth one-dimensional data array Ci by taking the difference between the third one-dimensional integral data array A′i and the fourth one-dimensional integral data array Bi (step S10). ). In this way, by removing the difference between the third one-dimensional integral data array A′i and the fourth one-dimensional integral data array Bi, excess noise that cannot be removed by the integral value correcting unit 116 is removed. can do.
Although the median filter is used here, a process such as moving average or Fourier transform may be used as long as it has a low-pass filter function. Similarly, unnecessary noise can be reduced by using the low-pass filter, and dust and dirt on the reference white plate and the reflection mirror can be accurately detected. Here, the low-pass filter is a filter that cuts the high-frequency region of a certain signal waveform and passes only the low-frequency region (= pass). Thus, the present invention can be applied to an image that can be subjected to Fourier transform, and unnecessary noise can be reduced by creating an image with a blurred outline by cutting high frequencies.
Further, when the integrated value correction unit 116 can sufficiently remove noise, instead of the median filter, the average value of the third one-dimensional integrated data array A′i is used instead of the fourth one-dimensional integrated data array Bi. You may use.
FIG. 7 is a graph showing the fifth one-dimensional data array Ci. In FIG. 7, it can be confirmed that the vertical streak component buried in the swell component in the stage of FIG.

FIG. 8 is a graph showing that a threshold value is provided to distinguish dirt and the like. A certain threshold value Th1 is provided from the waveform of FIG. 8, and a portion exceeding the threshold value Th1 can be taken as a candidate if dust or dirt is attached to the reference white plate or the reflecting mirror. Therefore, it can be seen that the undulation component is corrected to reveal only dirt such as dust. In the image evaluation apparatus 1 of the present invention, the illumination and the illuminance unevenness of the lens system are corrected even in the image data before shading correction. It can be corrected, and it can be detected with high accuracy as a candidate to which dust or dirt on the reference white plate or reflecting mirror is attached.
FIG. 9 is an enlarged view of the circled portion in FIG. The image evaluation value calculation unit 119 obtains an area surrounded by the waveform and the first threshold value Th1. Based on this area, the presence or absence of dust or dirt is determined. That is, if the area is larger than the second threshold value Th2, it is finally determined that there is dust or dirt, and if it is smaller, it is determined that there is no dust (steps S11 to S19). There is a risk of false detection due to noise if it simply exceeds a certain threshold, but by using such a feature quantity, it is possible to quantify the degree of abnormality as well as whether there is an abnormality. The risk of false positives can be reduced.
Further, if the area Gi surrounded by the threshold value of the pixel determined to be dust is larger than the second threshold value Th2 which is larger than the first threshold value Th1, the final state is when dust or dirt is present. If it is smaller, it is determined that it is smaller (steps S20 to S24). Here, the loop from step S20 to step S22 is repeated from the case where the pixel Xi is 1 to the case where j is sequentially j, and it is determined whether the evaluation result exceeds a predetermined threshold value for all the pixels (step S21). . If the evaluation result output unit 120 exceeds the threshold value, as a final determination, it is determined that there is dirt such as dust, and this loop is exited (step S24). However, if it is determined that the pixel Xi has not exceeded, and the pixel Xi has been determined to j (step S22), it is determined that there is no dirt such as dust as the final determination, and the series of processing ends (step S23). .
Then, the evaluation result output unit 120 externally outputs the evaluation result such as a monitor, a printer, or a file (step S25).

DESCRIPTION OF SYMBOLS 1 Image evaluation apparatus 11 Image processing apparatus 111 CPU
DESCRIPTION OF SYMBOLS 112 Memory 113 Image input part 114 Integral value calculation part 115 Approximation formula calculation part 116 Integral value correction part 117 Numerical value conversion part 118 Difference calculation part 119 Image evaluation value calculation part 120 Evaluation result output part 121 Image display part 12 Image reader

Japanese Patent Laid-Open No. 08-149249

Claims (5)

In an image evaluation apparatus for detecting dust and dirt on a reference white plate or a reflection mirror of a digital image reading apparatus,
The image evaluation device includes an integration value calculation unit that averages a read image of a reference white plate before shading correction for each pixel to generate a first one-dimensional integration data array;
An approximate expression calculation unit that approximates the first one-dimensional integral data array using a function of each pixel position;
An integration value correction unit that corrects the first one-dimensional integral data array using the reciprocal of the second one-dimensional data array calculated by the approximate expression calculation unit;
A numerical value conversion unit that numerically converts the third one-dimensional integral data array calculated by the integral value correction unit to generate a fourth one-dimensional integral data array;
A difference calculating unit that calculates a difference between the third one-dimensional integrated data array and the fourth one-dimensional integrated data array to generate a fifth one-dimensional data array;
A feature amount calculation unit for calculating a feature amount based on the fifth one-dimensional data array;
An image evaluation apparatus comprising: an evaluation result output unit configured to determine the presence or absence of dust or dirt on a reference white plate or a reflection mirror of the digital image reading device based on the feature amount calculated by the feature amount calculation unit. The image evaluation apparatus according to claim 1, wherein the numerical value conversion unit is a low-pass filter. In an image evaluation method for detecting dust and dirt on a reference white plate or a reflection mirror of a digital image reader,
An integration value calculation step of averaging a read image of the reference white plate before shading correction for each pixel to generate a first one-dimensional integration data array;
An approximate expression calculating step of approximating the first one-dimensional integral data array using a function of each pixel position;
An integration value correcting step of correcting the first one-dimensional integral data array using the reciprocal of the second one-dimensional data array calculated in the approximate expression calculating step;
A numerical value conversion step of generating a fourth one-dimensional integral data array by numerically converting the third one-dimensional integral data array calculated by the integral value correcting step;
A difference calculating step of calculating a difference between the third one-dimensional integrated data array and the fourth one-dimensional integrated data array to generate a fifth one-dimensional data array;
A feature amount calculating step of calculating a feature amount based on the fifth one-dimensional data array;
An image evaluation method comprising: determining whether dust or dirt is present on a reference white plate or a reflection mirror of a digital image reading device based on the feature amount calculated in the feature amount calculation step. The image evaluation method according to claim 3, wherein the numerical value conversion step is a low-pass filter.   A readable recording medium on which a program for realizing the image evaluation method according to claim 3 is recorded.

Images