JP5456404B2 - Defective pixel detection method and image processing apparatus - Google Patents

Description

  The present invention relates to a defective pixel detection method and an image processing apparatus for detecting defective pixels of a detector in an image capturing apparatus that captures an image of a subject.

  Radiographic imaging devices are used in various fields including, for example, medical diagnostic images and industrial nondestructive inspection. As a radiation detector for detecting radiation (X-rays, α-rays, β-rays, γ-rays, electron beams, ultraviolet rays, etc.) that has passed through a subject in a radiographic imaging apparatus, there is currently a flat that converts radiation into an electrical signal. Some use a panel type detector (FPD (Flat Panel Detector)).

  As a method of using FPD for a radiation image detector, for example, electron-hole pairs (e-h pairs) generated by a photoconductive film such as amorphous selenium by collecting radiation are collected and read out as an electrical signal. It has a direct system that directly converts it into an electrical signal and a phosphor layer (scintillator layer) that is made of phosphor that emits light (fluoresce) when incident radiation is applied. This phosphor layer converts radiation into visible light. There are two methods of reading out this visible light with a photoelectric conversion element, that is, an indirect method of converting radiation into an electric signal as visible light.

  Such an FPD has a structure in which a large number of radiation detection elements (pixels) for detecting radiation are arranged in a lattice pattern, and radiation is generated by using a switching element such as a thin film transistor (TFT). For example, the charges accumulated in the detection element are sequentially selected and output line by line.

In the radiographic image capturing apparatus using the FPD, defective pixels of the FPD are cited as one cause of the deterioration of the image quality of the radiographic image.
That is, the FPD pixel (radiation detection element) is not only suitable for outputting a signal having an appropriate intensity (concentration) with respect to the radiation dose that is always incident, but is also suitable for the radiation dose. There may be a defective pixel that outputs a signal with a value lower than the intensity or a signal with a higher value.
This defective pixel occurs due to various causes such as deterioration of the pixel due to irradiation of radiation, or a failure of an electric wiring for reading a signal such as a signal reading electrode or a switching element for reading a charge accumulated in the pixel. . For this reason, the degree of defect, that is, the intensity of the abnormal signal output from the defective pixel varies, and the shape of the defect is a point defect in which a single or a small number of defective pixels appear as dots, or a defective pixel is a line. There are various line defects such as line defects that appear in a line.

As a matter of course, a defective pixel cannot obtain a proper radiation image signal.
Further, the defective pixels of the FPD tend to increase as the number of radiographic image capturing increases.
Therefore, in the radiographic image capturing apparatus using the FPD, the position of the defective pixel of the FPD is detected at a predetermined timing, and when a radiographic image is captured, peripheral pixels ( Image defects are corrected using the image data), image correction is performed, and the image-corrected radiation image is displayed as a diagnostic image or the like and reproduced as a print.

  For example, Patent Document 1 discloses an image processing apparatus that detects a defective pixel of an FPD by separately detecting a point defect and a line defect of the FPD, and finally combining them. In this image processing apparatus, first, a defect detection image is acquired, and the first processing image obtained by performing median subtraction in the FPD reading direction on the defect detection image is used to make a point. By generating a defect extraction image and removing line defects from the point defect extraction image using defect continuity, etc., FPD point defects are detected, and offset correction and shading correction are performed on the defect detection image. The corrected image is subjected to median subtraction in the FPD reading direction and orthogonal direction to generate a line defect extraction image, and the defect defect continuity is used from the line defect extraction image. Is removed, and the line defect of the FPD is detected, and further, the defective pixel of the FPD is detected by synthesizing the detected point defect and line defect.

  Further, in Patent Document 2, in order to perform correction based on defect registration even for a defective pixel that is unstable over time, the defect detection frequency is counted for each pixel, and the defect detection frequency is determined as a defect setting value. An X-ray diagnostic apparatus is disclosed in which a symmetric pixel is registered as a defective pixel when it is larger than that, and an X-ray detection signal is calibrated based on the registration information of the defect. In this X-ray diagnostic apparatus, a subpixel consisting of a target pixel and a pixel adjacent to it is set for each of all pixels, and the value of each pixel in the subpixel region is obtained, and an intermediate value that is intermediate between them Is extracted by the median filter, and when the difference (absolute value) between the intermediate value extracted by the median filter and the value of the target pixel is larger than the pixel comparison set value, it is detected as an abnormality (missing) in the target pixel. .

JP 2008-252564 A JP 2007-143594 A

By the way, as also shown in Patent Document 1, the defective pixels of a detector such as an FPD include a line defect in which defective pixels are continuous in the pixel arrangement direction, and one or a plurality of pixels grouped into a defective pixel. There is a point defect that exists in Here, in the case of a line defect, a defective pixel outputs a signal having an abnormal intensity, so that an electrical influence is exerted on an adjacent normal line, and an abnormal output signal may be generated also in the adjacent pixel.
FIG. 7 is a graph showing an output signal of a part of an image in a direction orthogonal to the line defect in an image in which a line defect has occurred in an unexposed image obtained by reading data without irradiating the FPD with radiation. The horizontal axis represents the position, and the vertical axis represents the output signal (normalized signal). As shown in FIG. 7, the pixel in which the line defect occurs outputs an abnormally high value. It can be seen that the pixel on one side adjacent to the line defect outputs a lower value than the other normal pixels even though it is a normal pixel. In addition, it can be seen that a pixel having a shorter distance from the line defect outputs a lower value due to the influence of the line defect.

  A pixel whose output signal shows an abnormal value due to the influence of such a line defect (hereinafter referred to as “affected pixel”) is actually a normal pixel, and its output includes correct information.

  However, when a defect is detected using an image obtained by performing median subtraction with respect to the readout direction as a defect detection image as in Patent Document 1, or an intermediate value in a sub-pixel as in Patent Document 2. Is extracted by a median filter, and when the difference between the intermediate value and the value of the target pixel is larger than the pixel comparison setting value, the target pixel is detected as a defect. There is a possibility that the affected pixel is also detected as a defective pixel. Therefore, if defect correction of a captured image is performed based on these detection results, the affected pixels, that is, normal pixels are excessively corrected, and the correction accuracy is deteriorated.

Further, as described above, in Patent Document 1, after performing shading correction and offset correction on a defect detection image, median subtraction is performed in a reading direction and a direction orthogonal to the reading direction, thereby detecting the defect detection image. Is generated.
However, if shading correction or the like is performed on the read image, line defects and point defects are also corrected, and the defect range becomes smaller or the value becomes smaller than the actual defect. Therefore, when a defect is detected using this defect detection image, the defect range is detected to be smaller than the actual value, the value becomes smaller than the set threshold value, and the defect cannot be detected. If the threshold value is set low in order to detect this, the detection ability of the defect is deteriorated, for example, noise is erroneously detected as a defect.
Therefore, in order to detect the defects of the detector more reliably, it is preferable to use an uncorrected defect detection image that has not been corrected by shading correction or the like, but conversely without performing shading correction or the like. If defective pixel detection is performed, the risk of detecting the affected pixel as a defect increases.

  In this way, when a line defect affects an adjacent pixel and the adjacent pixel generates an abnormal output signal, the affected pixel is detected as a defect, or the affected pixel is detected by correction. Even if it is not detected as a defect, there is a possibility that the detection ability of the defect may be reduced, and it is impossible to perform high-precision defect correction on the photographed image and to provide a high-quality image. .

  An object of the present invention is to solve the problems of the prior art described above, where a line defect affects an adjacent pixel, and even when the adjacent pixel generates an abnormal output signal, a place where image correction is necessary. It is an object of the present invention to provide a defective pixel detection method and an image processing apparatus capable of accurately detecting the image and correcting the image accurately.

  In order to achieve the above object, the present invention provides a step of acquiring a defect detection image from the detector when detecting a defective pixel of a detector having two-dimensionally arranged pixels, and the defect detection The image is subjected to a one-dimensional smoothing process in a predetermined direction which is one direction of the pixel arrangement direction to generate a first smoothed image, and the first smoothed image is generated from the defect detection image. Subtracting to generate a first image, and performing a one-dimensional smoothing process on the first image in a direction orthogonal to the predetermined direction to generate a second smoothed image, Subtracting the second smoothed image from the first image to generate a second image; and detecting a point defect of the detector from the second image. A defective pixel detection method is provided.

Here, the defect detection image is preferably an uncorrected image.
The smoothing process is preferably a median process.
In the step of detecting a point defect from the second image, it is preferable that a pixel exceeding a predetermined threshold is detected as a point defect.
Furthermore, it is preferable to include a step of correcting a point defect of the detector.

The detector is preferably a flat panel radiation detector.
Here, it is preferable that the defect detection image is a uniform irradiation image obtained by uniformly irradiating the radiation detector with radiation and reading from the radiation detector.
Alternatively, the defect detection image is preferably an unexposed image read from the radiation detector without irradiating radiation.

  Further, in place of the step of generating the first image, the step of detecting a line defect of the detector, and the step of dividing the defect detection image at the position of the line defect, In the step of generating the image, a one-dimensional smoothing process is performed on each of the divided images in one direction of the pixel arrangement direction to generate a smoothed image, and each of the divided images It is preferable to generate the second image by subtracting the smoothed image corresponding to.

  The present invention also relates to an image processing device for processing an image output from a detector having two-dimensionally arranged pixels, and obtaining an image for defect detection from the detector at a predetermined timing. Then, a one-dimensional smoothing process is performed on the defect detection image in a predetermined direction which is one direction of the pixel arrangement direction to generate a first smoothed image, and the defect detection image A first image processing means for subtracting the first smoothed image to generate a first image; and a first-dimensional smoothing process is performed on the first image in a direction orthogonal to the predetermined direction. A second image processing means for generating a second smoothed image, subtracting the second smoothed image from the first image to generate a second image, and a point for detecting a point defect from the second image An image processing apparatus comprising: a defect detection unit; It is intended to provide.

Here, the defect detection image is preferably an uncorrected image.
The smoothing process performed by the first image processing unit and the second image processing unit is preferably a median process.
Further, it is preferable that the point defect detection means detects a pixel exceeding a predetermined threshold as a point defect.
Furthermore, it is preferable to have defect correction means for correcting defective pixels with respect to the image acquired from the detector.

The detector is preferably a flat panel radiation detector.
Here, it is preferable that the defect detection image is a uniform irradiation image obtained by irradiating the radiation detector with radiation uniformly and reading from the radiation detector.
Alternatively, the defect detection image is preferably an unexposed image read from the radiation detector without irradiating radiation.

  Further, instead of the first image processing means, it has a line defect detection means for detecting a line defect of the detector, and an image dividing means for dividing the defect detection image at the position of the line defect, The second image processing means performs a one-dimensional smoothing process on each of the divided images in one direction of the pixel arrangement direction to generate a smoothed image, and corresponds to each of the divided images Preferably, the smoothed image to be subtracted is used to generate the second image.

  According to the present invention, even when a line defect affects an adjacent pixel and the adjacent pixel generates an abnormal output signal, the image can be accurately corrected at a place where image correction is necessary, and high quality is achieved. Images can be provided.

It is a figure of one Embodiment which represents notionally the structure of the radiographic imaging apparatus to which this invention is applied. It is a block diagram showing the structure of the image process part of the radiographic imaging apparatus shown in FIG. (A)-(E) are the figures which show notionally the data for a defect detection for detecting a point defect. It is a flowchart for demonstrating the defect detection by the radiographic imaging apparatus shown in FIG. It is a block diagram showing the structure of another example of an image processing part. (A)-(D) are the figures which show notionally the data for a defect detection for detecting a point defect in the image processing part shown in FIG. It is a graph which shows the density | concentration of a part of image which a line defect generate | occur | produced.

  Hereinafter, a first aspect of a radiographic imaging apparatus to which the present invention is applied will be described in detail based on a preferred embodiment shown in the accompanying drawings.

  FIG. 1 is a diagram conceptually showing the configuration of a radiographic imaging apparatus to which an embodiment of the present invention is applied. A radiographic image capturing apparatus (hereinafter also referred to as an image capturing apparatus) 10 shown in FIG. 1 irradiates a subject (subject) H with radiation, and the radiation transmitted through the subject H is a radiation detector 32 of a radiation detection unit 30 described later. It detects and converts into an electric signal corresponding to image data, and generates a radiographic image in which the subject H is photographed based on the converted electric signal. The imaging device 10 includes an imaging unit 12, an imaging data processing unit 14, an image processing unit 16, an output unit 18, an imaging instruction unit 20, and a control unit 22.

The shooting instruction unit 20 is a part that sets a shooting menu, shooting conditions, and the like and instructs shooting of the subject H. The shooting instruction unit 20 is provided with a shooting menu, input keys for setting shooting conditions, and shooting instruction means.
In the example shown in the drawing, a two-push type photographing button is used as a photographing instruction means. When the shooting button is pressed down to the first level (half-pressed), it is ready for shooting, and when it is pressed down to the second level (fully pressed), shooting starts. The shooting instruction unit 20 includes information such as a shooting instruction signal indicating a state where the shooting button is not pressed, a state where the shooting button is pressed down to the first level, and a state where the shooting button is pressed down to the second level, and shooting conditions. Output to.

The control unit 22 is a part that controls the operation of each part of the imaging apparatus 10 in accordance with an imaging instruction signal or the like supplied from the imaging instruction unit 20.
For example, the control unit 22 controls the photographing unit 12 so that photographing is performed with the set photographing menu and photographing conditions.
As will be described in detail later, the imaging apparatus 10 detects a defect (defective pixel) of the radiation detector 32 and creates a defect map indicating information on the position and size of the defect in the radiation detector 32, and this defect map. The defect of the radiographic image image | photographed using is corrected. In order to detect the position of the defect, the control unit 22 controls the imaging data processing unit 14 to read out the defect detection data used for defect detection from the radiation detector 32 at the set timing. Specifically, the imaging apparatus 10 acquires defect detection data in order to detect the position of the defect when the imaging apparatus 10 is activated and at predetermined time intervals. At this time, defect detection data is acquired for each shooting mode described later.
Furthermore, the control unit 22 controls the image processing unit 16 so as to perform predetermined image processing on the captured radiation image (radiation image data).

  The imaging unit 12 includes a radiation source 26, an imaging table 28 for placing the subject H in a supine position at the time of imaging, and a radiation detection unit 30 that detects radiation transmitted through the subject H. In the imaging unit 12, the radiation source 26 emits radiation, and the radiation that has passed through the subject H on the imaging table 28 is detected by the radiation detector 32 of the radiation detection unit 30, thereby imaging the subject H. The imaging unit 12 outputs radiographic image data (analog data) obtained by imaging the subject H.

The radiation source 26 is a normal radiation source used for a radiographic imaging apparatus.
A radiation detector (Flat Panel Detector, hereinafter referred to as FPD) 32 detects radiation that has passed through the subject H, converts it into an electrical signal (radiation image data), and data of a radiation image obtained by photographing the subject H ( A detector of radiation (image) that outputs analog data), and converts incident radiation into an electrical signal by a photoconductive film of amorphous selenium.

  The photographing data processing unit 14 reads image data from the FPD 32 in accordance with an instruction from the control unit 22 and performs predetermined data processing such as A / D (analog / digital) conversion on the read image data. It is a part. The photographing data processing unit 14 outputs image data (digital data) after data processing.

The imaging data processing unit 14 basically reads image data from the FPD 32 according to the accumulation time. The accumulation time is a time during which radiation is converted into electric charges and accumulated in the FPD 32.
For example, when imaging the subject H, the imaging data processing unit 14 reads out a radiation image (radiation image data) from the FPD 32 after a predetermined accumulation time has elapsed after exposure.

  Here, the imaging data processing unit 14 reads out radiation images from the FPD 32 at different accumulation times according to the imaging menu. Therefore, the photographing apparatus 10 performs photographing mode division according to the accumulation time. In response to this, the imaging apparatus 10 acquires defect detection data for each imaging mode, detects defects, and creates a defect map. Note that the method for creating the defect map is the same regardless of the shooting mode, and therefore, in the following description, only one shooting mode will be basically described.

  That is, the imaging data processing unit 14 acquires defect detection data for detecting the position of the defect in the radiographic image for each imaging mode when the imaging apparatus 10 is activated and at predetermined time intervals. Specifically, the defect detection data is a uniform irradiation image (so-called solid image) that is data read from the FPD 32 by uniformly irradiating the FPD 32 with radiation.

  The defect detection data is a uniform irradiation image acquired by the imaging data processing unit 14 from the FPD 32. However, the present invention is not limited to this, and is data read from the FPD 32 without irradiation. An unexposed image may be used as defect detection data.

The image data read by the photographing data processing unit 14 is then sent to the image processing unit 16. As shown in FIG. 2, the image processing unit 16 includes an image acquisition unit 42 and an image correction unit 44, and offsets image data after data processing acquired from the imaging data processing unit 14. This is a part that performs predetermined image processing such as correction, shading correction, and defective pixel correction, and supplies the image data to the output unit 18 as output image data such as monitor display or print display. The image correction unit 44 also detects defects.
The image processing unit 16 is configured by a program (software) operating on a computer, dedicated hardware, or a combination of both. The image processing unit 16 outputs the radiographic image after image processing.

The image acquisition unit 42 acquires image data processed by the imaging data processing unit 14 in accordance with an instruction from the control unit 22 and supplies the image data to a predetermined part of the image processing unit 16.
As described above, since the imaging data processing unit 14 reads out the radiographic image obtained by imaging the subject H and the defect detection data for detecting defects from the FPD 32, the image acquisition unit 42 captures these image data. Obtained from the data processing unit 14 and supplied to a predetermined part.

The image correction unit 44 performs predetermined image processing on the image data (radiation image) acquired by the image acquisition unit 42 in accordance with an instruction from the control unit 22.
The image processing performed by the image correction unit 44 is not particularly limited, and data for line defect correction, offset correction, shading correction (gain correction), density correction, sharpness processing, gradation correction, monitor display, and print output Image processing (image correction) performed in various radiographic imaging devices and image processing devices, such as data conversion for converting image data, can be used. All of these corrections may be performed by a known method.
In the illustrated example, the image correction unit 44 also performs defect correction for correcting the point defect of the FPD 32, detects the position of the point defect, creates a defect map, and corrects the point defect for the radiation image data. A defect detection unit 46, a defect storage unit 48, and a defect correction unit 50.

  The image acquisition unit 42 supplies defect detection data to the defect detection unit 46 and a radiographic image of the subject H to the defect correction unit 50. In addition, the defect detection unit 46 supplies the defect detection result to the defect storage unit 48, and the defect correction unit 50 reads the defect information stored in the defect storage unit 48.

As described above, in the photographing apparatus 10, defective pixels connected to the FPD 32 in a line shape, that is, line defects may occur due to defective electrical wiring or the like. The line defect may affect the normal pixel line adjacent to the abnormal signal, and may generate an abnormal output in the adjacent pixel line. If a pixel whose output signal has an abnormal value due to the influence of the line defect (hereinafter referred to as “affected pixel”) is detected as a defective pixel, the accuracy of correction may be deteriorated.
The defect detection unit 46 is a part for detecting a defective pixel of the FPD 32 without detecting the affected pixel as a defect.

The defect detection unit 46 includes a first image processing unit 52, a second image processing unit 54, and a point defect detection unit 56, and defect detection acquired by the image acquisition unit 42 from the captured data processing unit 14. This is a part for detecting the position and size of the point defect of the radiographic image (that is, the position of the defective pixel) using the data for operation.
The defect detection unit 46 supplies the detected point defect information to the defect storage unit 48.

The first image processing unit 52 performs median subtraction in a predetermined direction on the defect detection data supplied from the image acquisition unit 42, and from the defect detection data, the line defect and the affected pixel extending in the predetermined direction. This is the part where the line is removed.
Specifically, the first image processing unit 52 adds the uncorrected defect detection data supplied from the image processing unit 42 in one direction of the pixel array (for example, when the reading electrode of the FPD 32 extends in one direction). In this direction as an example, a one-dimensional median filter is applied to generate a first smoothed image, and the first smoothed image is subtracted from uncorrected defect detection data to generate a first image.
The first image processing unit 52 supplies the generated first image to the second image processing unit 54.

The second image processing unit 54 performs median subtraction on the first image supplied from the first image processing unit 52 in a direction orthogonal to the predetermined direction in which the first image processing unit 54 performed median subtraction. From this, the line defect extending in the direction orthogonal to the predetermined direction and the line of the affected pixel are removed.
Specifically, the second image processing unit 54 is orthogonal to one direction of the image arrangement, which is the direction in which the first image processing unit 54 performs median subtraction on the first image supplied from the first image processing unit 52. A second smoothed image is generated by applying a one-dimensional median filter in a direction (for example, a direction in which the TFT of the FPD 32 is connected), and a second image is generated by subtracting the second smoothed image from the first image. To do.
The second image processing unit 54 supplies the generated second image to the point defect detection unit 56.

Here, the generation of the point defect detection image (second image) will be described more specifically with reference to a diagram conceptually showing the defect detection data in FIG. FIG. 3A is a schematic diagram of uncorrected defect detection data acquired by the imaging data processing unit 14 from the FPD 32, and has line defects in one direction of the pixel array and a direction perpendicular thereto, and the entire image is displayed. Has distributed point defects. 3A to 3E, the horizontal direction in the figure is the x direction, and the vertical direction is the y direction.
As described above, the imaging apparatus 10 acquires defect detection data at the time of activation and at a predetermined time interval. For example, the first image processing unit 52 of the defect detection unit 46 applies a one-dimensional median filter in the x direction to the acquired defect detection data. The line filter extending in the y direction and the point defect are removed from the defect detection data by the processing by the media filter, and the line defect extending in the x direction and the affected pixel line and the y direction are extended. A first smoothed image (FIG. 3B) in which the affected pixel line remains is generated.

Here, when the median filter is applied, the affected pixel lines extending in the direction orthogonal to the median filter direction remain, as shown in FIG. This is because there is a width of ten pixels, so even if the median filter is applied, the value of the affected pixel is reduced but not completely removed. If the median filter is sufficiently larger than the width of the affected pixel line, the affected pixel line is also completely removed. A median filter sufficiently larger than the line width of the affected pixel can also be applied to the present invention.
Further, depending on the median filter setting, the value of the pixel of the line defect extending in the direction orthogonal to the direction of the median filter after the median filter is applied is not equal to the value of the normal pixel, and is not equal to the normal pixel value. In some cases, the level is equal to the value of the affected pixel. Such a median filter can also be applied to the present invention.

Further, the first image processing unit 52 subtracts the first smoothed image from the defect detection data, and the line defect and the affected pixel line extending in the x direction are removed from the defect detection data. A first image shown in FIG. 3C is generated and supplied to the second image processing unit 54.
Since the affected pixel line extending in the y direction of the first smoothed image has a small value, even if the first smoothed image is subtracted from the defect detection data, the line extends in the y direction. The affected pixel line is not completely removed, and the affected pixel line whose value has decreased remains.

Next, the second image processing unit 54 applies a one-dimensional median filter to the first image supplied from the first image processing unit 52 in the y direction of the image. By the processing by the median filter, the point defect is removed, and a second smoothed image (FIG. 3D) in which the line defect extending in the y direction and the affected pixel line remain is generated.
Further, the second image processing unit 54 subtracts the second smoothed image from the first image to remove the line defect and the affected pixel line extending in the y direction from the first image, and only the point defect is obtained. A second image shown in FIG. 3E is generated and supplied to the point defect detection unit 56.

  Note that the median subtraction directions performed by the first image processing unit 52 and the second image processing unit 54 are not limited to the x direction and the y direction, respectively. The first image processing unit 52 performs the y direction and the second image. The processing unit 54 may perform median subtraction in the x direction.

  Further, the first image processing unit 52 and the second image processing unit 54 generate the first and second smoothed images by applying a median filter to the image, respectively, but the present invention is not limited to this. Alternatively, various smoothing filters such as a moving average filter, a weighted average filter, or a low-pass filter can be used. Even in these cases, the first and second smoothed images are generated using the one-dimensional smoothing filter.

The point defect detection unit 56 is a part that detects a point defect of the radiation image (FPD 32) from the second image supplied from the second image processing unit 54.
In the present invention, as an example, the value of each pixel in the second image is compared with a predetermined threshold, and a pixel that exceeds the predetermined threshold is detected as a defective pixel (point defect), and the position of the defective pixel in the FPD 32 is indicated. Create a defect map.
The point defect detection unit 56 supplies the detected defect map to the defect storage unit 48.

  The point defect detection method performed by the point defect detection unit 56 is not particularly limited, and all point defect detection methods performed in various image processing apparatuses can be used.

As described above, in the photographing apparatus 10, when a line defect occurs in the FPD 32, the line defect may affect the line of adjacent pixels, and an affected pixel may be generated.
On the other hand, in the imaging apparatus 10, the first image processing unit 52 performs median subtraction on the defect detection data in one direction of the pixel array, and the line defect extending in one direction of the pixel array and the influence thereof. The pixel line is removed from the defect detection data. Further, the second image processing unit 54 performs median subtraction to remove line defects and affected pixel lines extending in a direction orthogonal to one direction of the previous pixel array. The point defect detection unit 56 detects point defects using an image (second image) from which line defects in one direction of the pixel array and a direction perpendicular thereto are removed from the defect detection data (second image). Therefore, it is possible to detect the defect with high accuracy without erroneously detecting the affected pixel as a defective pixel.

The defect storage unit 48 is a part that stores the defect map supplied from the defect detection unit 46.
Further, the defect storage unit 48 supplies the defect map to the defect correction unit 50 in accordance with the instruction.

The defect correction unit 50 captures the subject H, reads it with the FPD 32, and corrects the defect according to the defect map read from the defect storage unit 48 with respect to the radiation image supplied from the image acquisition unit 42. is there.
The defect correction method is not particularly limited. Based on the read defect map, normal pixel data in the vicinity of the defect is used as image data of the defective pixel, and the average value of the pixels adjacent to and around the defect is calculated. An image defect correction method performed in various types of radiographic imaging apparatuses, such as a method for obtaining defective pixel data, can be used.
The defect correction unit 50 supplies the radiation image with the defect corrected to the next step, in the illustrated example, to the output unit 18.

  The output unit 18 is a part that outputs a radiographic image after image processing supplied from the image processing unit 16. The output unit 18 includes, for example, a monitor that displays a radiation image on a screen, a printer that prints out a radiation image, and a storage device that stores radiation image data.

Next, referring to FIG. 1 and FIG. 2, the operation of photographing in the photographing apparatus 10 will be described.
The radiographic image capturing of the imaging apparatus 10 according to the present invention is the same as a known radiographic image capturing apparatus. That is, when imaging is instructed by the imaging instruction unit 20, the radiation source 26 irradiates the subject H with radiation, and the FPD 32 detects the radiation transmitted through the subject H and converts it into an electrical signal. A radiographic image is generated. The imaging data processing unit 14 reads the generated radiographic image, performs predetermined data processing such as A / D (analog / digital) conversion on the read radiographic image, and obtains the radiographic image after data processing as an image acquisition unit. 42 is acquired and supplied to the image processing unit 16.
The image processing unit 16 performs necessary image processing such as shading correction and offset correction, and performs defect correction on the radiation image based on the defect map stored in the defect storage unit 48, thereby correcting the defect. The subsequent radiation image is supplied to the output unit 18. The output unit 18 performs monitor display of the acquired radiation image, print output from a printer, storage in a storage device, and the like.

  Here, the defect correction will be described with reference to the flowchart of FIG. The imaging apparatus 10 acquires defect detection data for each mode at startup and at predetermined time intervals. The defect detection unit 46 detects a point defect from the acquired defect detection data, creates a defect map, and supplies the defect map to the defect storage unit 48.

The defect storage unit 48 stores the defect map supplied from the defect detection unit 46 for each mode.
When the radiographic image of the subject is captured, the defect correction unit 50 selects one of the defect maps stored in the defect storage unit 48 according to the imaging mode, reads it, and uses the defect map as a basis. Then, the defect of the captured radiographic image is corrected, and the radiographic image after the defect correction is supplied to the output unit 18.

  As described above, the line defect may affect the normal pixel line adjacent to the abnormal signal, and may generate an abnormal output in the adjacent normal pixel line. If a defective pixel is detected using an image having an affected pixel affected by a line defect, the defective pixel cannot be correctly detected, such as detecting the affected pixel as a defective pixel, and the correction accuracy is improved. May be worse.

  On the other hand, according to the first aspect of the present invention, as described above, median subtraction is performed on the defect detection data in one direction of the pixel array, and further in a direction orthogonal to the one direction of the pixel array. Performing median subtraction to detect a defective pixel using an image from which the affected pixel line extending in one direction of the pixel array and the orthogonal direction is removed, so that the affected pixel is erroneously detected as a defective pixel. There is no.

In the above-described example, defective pixels are detected using defect detection data output from the imaging data processing unit 14 without performing image processing such as shading correction. However, the present invention is not limited to this. Instead, defective pixels may be detected using defect detection data subjected to shading correction, offset correction, or the like. However, if correction such as shading correction is performed on the defect detection data, the point defect range may be reduced or the value may be reduced depending on the type of defect. If the threshold value is lowered in order to detect a point defect whose size has become smaller, noise may be erroneously detected as a defect, and the point defect detection capability will deteriorate.
Therefore, in order to detect the defective pixels of the FPD more accurately and reliably, it is preferable to detect point defects using uncorrected defect detection data. However, if a defect is detected by a conventional defective pixel detection method using uncorrected defect detection data, an affected pixel that is a normal pixel is also detected as a defective pixel.

  On the other hand, according to the present invention, even if point defect detection is performed using uncorrected defect detection data that has not been subjected to correction such as shading correction, an affected pixel is erroneously detected as a defective pixel. In addition, since a point defect can be detected, a defective pixel included in the FPD can be accurately and reliably detected without erroneous detection. Therefore, according to the present invention, it is possible to detect defective pixels with high accuracy, accurately detect a portion requiring image correction, accurately correct the image, and capture a high-quality radiographic image. Become.

  Further, in the above-described embodiment, the imaging mode is divided according to the accumulation time. However, the present invention is not limited to this, and the mode may be divided according to the radiation dose at the time of photographing the subject. The mode may be divided according to the combination of time and dose.

  In the first aspect of the present invention described above, the median subtraction is performed on the defect detection data in one direction of the pixel array, and further, the median subtraction is performed in a direction orthogonal to the one direction of the pixel array. A point defect is detected using an image in which a line of affected pixels extending in one direction of the array and the direction orthogonal thereto is deleted.

  On the other hand, in the second aspect of the present invention, instead of the one-way median subtraction, the defect detection data is divided based on the information of the line defect detected in advance, and the divided defect detection data is obtained. On the other hand, point defects are detected using an image obtained by performing median subtraction only in one direction.

FIG. 5 shows a block diagram of an example of the image processing unit 80 in the radiographic imaging apparatus that detects such point defects.
As an example, the second aspect of the radiographic image capturing apparatus of the present invention has a configuration in which an image processing unit 80 shown in FIG. 5 is arranged in place of the image processing unit 16 in the imaging apparatus 10 shown in FIG.
The image processing unit 80 shown in FIG. 5 has the same configuration as the image processing unit 16 except that it includes a line defect detection unit 88 and a third image processing unit 86 instead of the first image processing unit 52. Therefore, the same parts are denoted by the same reference numerals, and the following description will mainly be made on different parts.

The line defect detection unit 88 is a part that detects a line defect in the radiation image (FPD 32).
The line defect detection method performed by the line defect detection unit 88 is not particularly limited, and the line defect may be detected from the image data supplied from the image acquisition unit 42, or the operator inputs with a mouse or a keyboard. Information on line defects may be acquired.
The line defect detection unit 88 stores information on the detected line defect and supplies the information to the third image processing unit 86.

  The line defect detection method when the line defect detection unit 88 detects a line defect from image data is not particularly limited. For example, the line defect may be detected with a threshold. As shown, median subtraction may be performed to detect line defects. The image data used for defect detection may be defect detection data used for point defect detection, or image data (line defect detection data) different from the defect detection data used for point defect detection. There may be.

The third image processing unit 86 is a part that divides the defect detection data supplied from the image acquisition unit 42 based on the line defect information supplied from the line defect detection unit 88.
The third image processing unit 86 divides the uncorrected defect detection data supplied from the image processing unit 42 into pixels having line defects based on the line defect information supplied from the line defect detection unit 88. Then, a third image composed of a plurality of images is generated.
The third image processing unit 86 supplies the generated third image to the second image processing unit 54.

  Detection of point defects in the image processing unit 80 will be described with reference to a diagram conceptually showing the defect detection data shown in FIG. FIG. 6A is a schematic diagram of uncorrected defect detection data acquired by the imaging data processing unit 14 from the FPD 32, and has a line defect in one direction of the pixel array and a direction orthogonal thereto, and the entire image is displayed. Has distributed point defects. 6A to 6D, the horizontal direction in the figure is the x direction and the vertical direction is the y direction.

  Similarly to the first aspect, the imaging apparatus 10 acquires defect detection data for each mode at the time of activation and at predetermined time intervals. The defect storage unit 48 stores information on the line defects acquired in advance. The third image processing unit 86 of the defect detection unit 84 divides the acquired defect detection data by the pixels where the line defect exists based on the information on the line defect supplied from the line defect detection unit 88, In the illustrated example, a third image composed of four images is generated (FIG. 6B). Since the third image divides the image at the position of the line defect, the line of the affected pixel that appears adjacent to the line defect exists at the end of each divided image in the third image.

Next, the second image processing unit 54 applies a one-dimensional median filter to the third image supplied from the third image processing unit 86 in the y-direction of the image, and a third smoothed image configured by a plurality of images. Is generated. Although the point defect is removed by the median filter, the line of the affected pixel is not removed by the median filter because the line of the affected pixel exists at the edge of the image. Therefore, the third smoothed image is a plurality of (four) images in which only the lines of the affected pixels remain as shown in FIG.
Further, the second image processing unit 54 subtracts the corresponding third smoothed image from the third image to remove the affected pixel line from the third image, and only the point defect remains, FIG. ) And is supplied to the point defect detection unit 56.

The point defect detection unit 56 detects a point defect using the fourth image supplied from the second image processing unit 54 and creates a defect map.
The defect storage unit 48 stores the defect map supplied from the point defect detection unit 56 for each mode.

  The direction of median subtraction performed by the second image processing unit 54 may be either the x direction or the y direction. For example, when an image has many line defects in the y direction due to the reading direction or the like, it is preferable to perform median subtraction in the y direction. Specifically, since many line defects (and affected pixels) occur in the connection direction of the signal reading electrode for reading the charge, it is preferable to perform median subtraction in the connection direction of the signal reading electrode.

As described above, according to the second aspect of the present invention, the defect detection data is divided based on the information of the line defect acquired in advance, and the median subtraction is performed in one direction of the pixel array, thereby affecting the affected pixels. Since the point defect is detected using the image from which the line is removed, the affected pixel is not erroneously detected as a defective pixel.
As in the first aspect, in order to detect defective pixels with high accuracy, it is preferable to detect point defects using uncorrected defect detection data that has not been subjected to correction such as shading correction. Even when the point defect is detected using the defect detection data for correction, the point defect can be detected without erroneously detecting the affected pixel as a defective pixel. Therefore, it is possible to detect defective pixels with high accuracy, accurately detect a portion that requires image correction, accurately correct the image, and capture a high-quality radiographic image.

  In addition, in the second aspect, although it is necessary to obtain information on line defects in advance, an image for detecting point defects can be generated by median subtraction only in one direction. Compared to the first mode in which the median subtraction is performed, the processing speed is increased.

  The radiographic imaging apparatus to which the defective pixel detection method of the present invention is applied has been described in detail above. However, the present invention is not limited to the above-described embodiments, and various improvements and modifications can be made without departing from the gist of the present invention. Of course, changes may be made. Of course, the image processing apparatus and the defective pixel detection method of the present invention can also be applied to an image processing apparatus that processes an image captured using a detector having two-dimensionally arranged pixels. .

10. Imaging device (radiological imaging device)
DESCRIPTION OF SYMBOLS 12 Imaging | photography part 14 Imaging | photography data processing part 16, 80 Image processing part 18 Output part 20 Imaging | photography instruction | indication part 22 Control part 26 Radiation source 28 Imaging stand 30 Radiation detection part 32 FPD (radiation detector)
42 Image acquisition unit 44, 82 Image correction unit 46, 84 Defect detection unit 48 Defect storage unit 50 Defect correction unit 52 First image processing unit 54 Second image processing unit 56 Point defect detection unit 86 Third image processing unit 88 Line defect Detection unit H Subject

Claims (16)

In detecting a defective pixel of a detector having two-dimensionally arranged pixels,
Obtaining a defect detection image from the detector;
Detecting line defects in the detector;
Dividing the defect detection image at the position of the line defect;
A one-dimensional smoothing process is performed on each of the divided images in one direction of the pixel arrangement direction to generate a smoothed image, and the smoothed image corresponding to each of the divided images is subtracted. And generating a second image;
And detecting a point defect of the detector from the second image.   The defective pixel detection method according to claim 1, wherein the defect detection image is an uncorrected image.   The defective pixel detection method according to claim 1, wherein the smoothing process is a median process.   The defective pixel detection method according to claim 1, wherein the step of detecting a point defect from the second image detects a pixel that exceeds a predetermined threshold as a point defect.   Furthermore, the defective pixel detection method in any one of Claims 1-4 which has the process of correct | amending the point defect of the said detector.   The defective pixel detection method according to claim 1, wherein the detector is a flat panel type radiation detector.   The defective pixel detection method according to claim 6, wherein the defect detection image is a uniform irradiation image obtained by uniformly irradiating the radiation detector with radiation and reading from the radiation detector.   The defective pixel detection method according to claim 6, wherein the defect detection image is an unexposed image read from the radiation detector without irradiating radiation. In an image processing apparatus for processing an image output from a detector having two-dimensionally arranged pixels,
Image acquisition means for acquiring a defect detection image from the detector at a predetermined timing;
A line defect detecting means for detecting a line defect of the detector;
Image dividing means for dividing the defect detection image at the position of the line defect;
A smoothed image is generated by performing a one-dimensional smoothing process on each image divided by the image dividing means in one direction of the pixel arrangement direction, and a smoothing corresponding to each divided image. A second image processing means for subtracting the digitized image to generate a second image;
An image processing apparatus comprising point defect detection means for detecting a point defect from the second image.   The image processing apparatus according to claim 9, wherein the defect detection image is an uncorrected image. Before Stories second image processing means smoothing process performed is the image processing apparatus according to claim 9 or 10 is a median processing.   The image processing apparatus according to claim 9, wherein the point defect detection unit detects a pixel that exceeds a predetermined threshold as a point defect.   The image processing apparatus according to claim 9, further comprising defect correction means that corrects defective pixels with respect to an image acquired from the detector.   The image processing apparatus according to claim 9, wherein the detector is a flat panel radiation detector.   The image processing apparatus according to claim 14, wherein the defect detection image is a uniform irradiation image obtained by uniformly irradiating the radiation detector with radiation and reading the radiation detector.   The image processing apparatus according to claim 14, wherein the defect detection image is an unexposed image read from the radiation detector without irradiating radiation.

Images