Radiation image signal processing method, radiation image signal processing device, radiation imaging system, and program

Abstract

The present invention performs image signal processing capable of implementing efficient data compression with suppressing degradation of data acquired in imaging using radiation. A radiation imaging system (100) has an image sensor (105) with a sensitive surface on which a plurality of pixels are two-dimensionally arrayed, and outputs, as a pixel value, a signal electrically converted in the pixels in accordance with incident of radiation containing image information. In the radiation imaging system (100), one or a plurality of pixels that output a significant electrically-converted signal originating from the incident radiation and a pixel that outputs only a signal originating from noise specific to a sensor device are separated using a threshold value statistically predetermined as a reference. Then, a pixel value originating from the noise is eliminated and a pixel value of the significant signal is output as radiation image data.

Description

[Technical Field]

[0001] The present invention relates to a radiation image signal processing method, a radiation image signal processing device, a radiation imaging system, and a program, and more particularly to a radiation image signal processing method, a radiation image signal processing device, a radiation imaging system, and a program for converting high-speed, wide-band image signals read out from an image sensor for imaging. [Background technology]

[0002] Radiation imaging systems that utilize radiation with strong penetrating power, such as X-rays, are used in fields such as material structure analysis and non-destructive testing of microfabricated products such as semiconductor devices. Such radiation imaging systems generally comprise a radiation source such as X-rays, an image sensor that receives radiation transmitted through or diffracted and scattered by a sample and converts it into an image signal, and a signal processing system that processes the image signal to obtain an analysis image. From the perspective of improving analytical accuracy and expanding use, it is important that the performance of all components is consistent. In particular, in line with the development of large, high-intensity radiation sources such as X-ray free electron lasers (abbreviated as XFEL), systems with megapixel-order pixel arrays and capable of capturing 1 to 10 images per second are being developed. 8 High-performance image sensors are being developed that have a wide dynamic range capable of detecting individual radiation particles and that can handle frame rates on the order of kHz. To avoid the so-called pile-up phenomenon, the preferred image sensors in this field are charge-integrating sensors with a complementary metal-oxide-semiconductor (CMOS) structure that accumulate charges generated in response to incident radiation particles over a fixed exposure time and output a corresponding image signal.

[0003] On the other hand, in high-performance radiation imaging systems that use the above-mentioned XFEL as a radiation source, large amounts of data are read from the image sensor when observing chemical changes inside materials, etc. Incidentally, in an X-ray imaging system that uses an XFEL emitting 10 KeV X-ray pulse laser light as a light source and has a camera system composed of a megapixel image sensor, it is necessary to process high-speed data signals of several GB / s captured at a frame rate of several KHz. If this is done as is, the data storage device and the computer system for signal processing would become extremely large and would be impractical. Therefore, data compression technology is an essential requirement from the perspective of preventing such systems from becoming too large.

[0004] On the other hand, the huge amount of data read from the image sensor contains noise components specific to the semiconductor devices that make up the sensor device. Therefore, in order to efficiently compress image data, it is preferable to remove pixel data representing pixel signal values ​​caused only by noise from the raw data group read from each pixel and convert the data into a form suitable for compression. From this perspective, methods for removing noise components before data compression have been proposed, as described in the following prior art documents. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] International Publication No. 2019 / 064632 [Non-patent literature]

[0006] [Non-Patent Document 1] M. Hammer et al., “Strategies for on-chip digital data compression for X-ray pixel detectors”, Journal of Instrumentation, Volume 16, January 2021, P01025 [Non-patent document 2] Ch. Broennimann et al., “The PILATUS 1M detector”, 2006, Journal of Synchrotron Radiation, Volume 13, Issue 2, pp.120-130 [Non-Patent Document 3] V. Radicci et al., “The International School for Advanced Studies (SISSA), find out more EIGER a new single photon counting detector for X-ray applications: performance of the chip”, 2012, Journal of Instrumentation, Volume 7, C02019 [Non-Patent Document 4] J. Hoff et al., “An on-chip charge cluster reconstruction technique in the miniVIPIC pixel readout chip for X-ray counting and timing”, Proceedings in 2014 IEEE, Nuclear Science Symposium and Medical Imaging Conference (NSS / MIC), 1173-1183 [Non-Patent Document 5] S. Cartier et al., “Micrometer-resolution imaging using MONCH: towards G2-less grating interferometry”, 2016, Journal of Synchrotron Radiation 23, pp.1462-1473 [Non-Patent Document 6] S. Cartier, “Development of a 25 micron pixel detector for phase-contrast imaging”, 2017, PhD thesis [Non-Patent Document 7] GW Deptuch et al., “An Algorithm of an X-ray Hit Allocation to a Single Pixel in a Cluster and Its Test-Circuit Implementation”, 2018, IEEE Transactions on Circuits and Systems I (TCSI), Volume 65, Issue 1, pp.185-197 [Non-patent document 8] J. Jirsa et al., “Simulation of New Charge Summing and Hit Allocation Algorithm”, Proceedings of Science, Topical Workshop on Electronics for Particle Physics [Non-Patent Document 9] H. Spieler, “Semiconductor Detector Systems”, 2005, Oxford University Press, Section 2.2.2 & 2.7 Summary of the Invention [Problem to be solved by the invention]

[0007] Data compression is an effective way to efficiently transfer and store the wideband data obtained from such high-quality image sensors, but it is necessary to compress the data without degrading the data derived from the received radiation as much as possible and in a way that removes the noise components inherent to the image sensor. Specifically, data compression is performed by removing noise from the data read out from the image sensor. However, when using high-energy rays such as the X-rays from the XFEL mentioned above, a phenomenon known as charge coupling occurs in the neighboring pixels around the pixel that received the radiation, producing a relatively weak signal that is difficult to distinguish from noise.

[0008] In light of this situation, various methods have been proposed to remove noise by clearly distinguishing between noise and relatively weak signals that spread near pixels where radiation is incident due to the incidence of high-energy rays. For example, a method has been proposed in which the pixel value (signal intensity) of each pixel is compared with a threshold value, and if the pixel value is smaller than the threshold value, the pixel value is determined to be noise and removed (Non-Patent Documents 1 to 3). In addition, a method has been proposed in which pixel values ​​of an area containing multiple pixels are discriminated using a threshold value (Non-Patent Documents 4 to 8).

[0009] However, these general methods mechanically determine whether a pixel value corresponds to noise based on its magnitude relative to a uniquely determined threshold, and therefore, depending on the setting of the threshold, significant pixel values ​​that are not noise may be removed, or noise components may be detected as false information, which may undermine the reliability of the detection signal. That is, in the field of radiation measurement technology, which performs structural analysis of materials at the atomic or molecular level or observation of chemical reactions, it is extremely important to accurately grasp information about radiation that has passed through or scattered through the observed sample in units of radiation photons or particles. If significant pixel values ​​that appear in pixels surrounding a pixel where radiation is incident due to the charge coupling phenomenon are removed as they are mixed in with noise, not only will this degrade the quality of the image, but it will also reduce the reliability of the analysis results.

[0010] The present invention has been made in consideration of the above circumstances, and has as its object to perform image signal processing that can achieve efficient data compression while suppressing degradation of data obtained in imaging using radiation. [Means for solving the problem]

[0011] A radiation image signal processing method according to one embodiment is characterized in that, in a radiation imaging system including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and outputting signals electrically converted by the pixels in response to incidence of radiation containing image information as pixel values, the method comprises the steps of: statistically separating one or more pixels that output significant electrically converted signals derived from the incident radiation from pixels that output only signals due to noise inherent in the sensor device using a predetermined threshold value; removing pixel values ​​due to noise; and outputting pixel values ​​of significant signals as radiation image data. This allows for the pixel values ​​due to noise to be suitably removed while maintaining image information, thereby obtaining image data that can be highly compressed.

[0012] A radiation image signal processing device according to one embodiment is a radiation imaging system including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and which outputs signals electrically converted by the pixels in response to incident radiation as image information as pixel values, the radiation image signal processing device comprising: a constant setting unit that determines a threshold value for detecting noise components appearing in each pixel; and a threshold discriminator that discriminates between pixels that output significant electrically converted signals derived from the incident radiation based on the threshold value set by the constant setting unit and pixels that output only signals derived from noise inherent in the sensor device, the constant setting unit being characterized in that a signal obtained by statistically processing previously acquired data is set as the threshold level in the threshold discriminator. This makes it possible to preferably remove pixel values ​​derived from noise while maintaining image information, thereby obtaining image data that can be highly compressed.

[0013] A radiation imaging system according to one embodiment includes a radiation source that irradiates radiation onto an observation sample, a camera unit including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and an image signal processing unit that outputs, as pixel values, signals electrically converted by the pixels in response to incidence of the radiation containing information about the observation sample, the image signal processing unit including a constant setting unit that determines a threshold for detecting noise components appearing in each pixel, and a threshold discriminator that discriminates between pixels that output significant electrically converted signals derived from radiation based on the threshold set by the constant setting unit and pixels that output only signals derived from noise inherent in a sensor device, the constant setting unit being characterized in that a signal obtained by statistically processing previously acquired data is set as the threshold in the threshold discriminator. This makes it possible to obtain image data that can be highly compressed by suitably removing pixel values ​​derived from noise while maintaining image information.

[0014] According to one embodiment, in a radiation imaging system including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and outputting signals electrically converted by the pixels in response to incidence of radiation containing image information as pixel values, the program causes a computer included in the radiation imaging system to execute the following processes: a process of statistically separating, based on a predetermined threshold, one or more pixels that output significant electrically converted signals derived from the incident radiation from pixels that output only signals due to noise inherent in the sensor device; and a process of removing pixel values ​​due to noise and outputting pixel values ​​of significant signals as radiation image data. This makes it possible to obtain image data that can be highly compressed by preferably removing pixel values ​​due to noise while maintaining image information. [Effects of the Invention]

[0015] According to the present invention, it is possible to perform image signal processing that can achieve effective data compression while suppressing degradation of data obtained in imaging using radiation. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a diagram schematically illustrating the configuration of an X-ray imaging system. [Figure 2] FIG. 2 is a diagram schematically illustrating a partial configuration of a sensitive surface of an image sensor that is an X-ray detector. [Figure 3] 1 is a diagram schematically showing an example of pixel values ​​that appear on pixels on the sensitive surface 10 when X-ray photons are incident thereon. [Figure 4] 2 is a diagram schematically illustrating the functional configuration of an image signal processing unit 140. FIG. [Figure 5] 4 is a flowchart of a noise removal process according to the first embodiment. [Figure 6] FIG. 4 is a diagram showing the order of sweeping a processing target area by a block of four pixels in the noise removal processing according to the first embodiment. [Figure 7] 1A and 1B are diagrams showing the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal processing using general threshold discrimination. [Figure 8] 10A and 10B are diagrams illustrating the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal processing using the 2×2-sum method. [Figure 9] 10 is a flowchart of a noise removal process according to the second embodiment. [Figure 10] FIG. 10 is a diagram showing the order of pixel sweep in the noise removal processing according to the second embodiment. [Figure 11] 1A and 1B are diagrams showing the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal by general threshold discrimination. [Figure 12] 10A and 10B are diagrams showing the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal processing by the T2C method. [Figure 13] 11 is a flowchart of a noise removal process according to the third embodiment. [Figure 14] FIG. 1 illustrates an example of the configuration of a computer. DETAILED DESCRIPTION OF THE INVENTION

[0017] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same elements are designated by the same reference numerals, and redundant explanations will be omitted as necessary.

[0018] Briefly, the present invention, which will be described below, is based on the idea of ​​separating one or more pixels that output a significant electrical conversion signal derived from radiation incident on the sensitive surface of an image sensor from pixels that output only signals due to noise inherent to the sensor device based on a statistically determined threshold, and generating a group of true radiation image data suitable for data compression by removing pixel values ​​due to noise. The present invention is also characterized by the idea of ​​compensating for the loss of significant information due to noise removal by adding a certain correction constant to one or multiple adjacent pixel values ​​that output a significant signal in response to incident radiation so that the sum of the pixel values ​​falls within a statistically determined range. Here, the statistically determined threshold and correction constant are dynamically set as individual values ​​corresponding to each observation event.

[0019] More specifically, the present invention has a first feature in which pixel signal values ​​from an image sensor having a pixel array-type sensitive surface that receives radiation and photoelectrically converts it are summed to obtain a sum signal, and then the sum signal is sequentially compared with a predetermined threshold to remove noise-caused pixel values ​​on a block-by-block basis. In the present invention, this noise removal technique according to the first feature is referred to as the 2×2-SUM (Two by Two Sum) method. The output signal from each pixel of the image sensor is compared with a first threshold determined based on a predetermined noise level to obtain an image signal from which noise components have been removed. This image signal is then compared again with a second threshold lower than the first threshold, and a correction constant is added to pixel values ​​detected between the first and second thresholds to generate an image signal. Image data corresponding to the incident radiation information is obtained by performing the two thresholds and the correction summation. In the present invention, this noise removal technique according to the second feature is referred to as the T2C (Threshold to Correction) method.

[0020] The 2x2-SUM method according to the first feature and the T2C method according to the second feature have a common technical significance in that they solve the problem of the prior art in that, when trying to accurately grasp pixel values ​​that represent significant information, pixel values ​​of noise components are included in the image data group, and when trying to completely remove the pixel values ​​of the noise components, the significant information is lost. Furthermore, in both methods, the threshold and correction constants used as discrimination criteria are obtained by statistically sorting a large amount of image data under various conditions such as the intensity of the radiation source, the image sensor and its exposure conditions, and the reference sample, and therefore, they are closely related to each other technically.

[0021] The present invention also includes a radiation image signal processing device having, as hardware, a signal processing function for executing the radiation image signal processing method, a radiation imaging system, and a software program for executing the image signal processing method of the present invention on a computer. Note that, although the radiation in the present invention includes all types of radiation such as electron beams, X-rays, gamma rays, ultraviolet rays, neutron beams, and charged particle beams, the following description will focus on X-rays as a representative example.

[0022] Embodiment 1 (X-ray imaging system) 1 shows a schematic configuration of an X-ray imaging system 100 according to the first embodiment. The X-ray imaging system 100 includes, as basic components of a direct imaging method, an X-ray source 101, an X-ray attenuation plate 102, a sample holder 104 for holding a sample 103 to be imaged, and an image sensor 105 constituting a camera unit arranged with its optical axis aligned with the X-rays irradiated from the X-ray source.

[0023] The X-ray source 101 is configured as, for example, an X-ray free electron laser (XFEL) and generates pulsed coherent X-rays with high energy of 10 keV. The X-rays emitted from the X-ray source 101 pass through the sample 103 and are attenuated appropriately by the X-ray attenuation plate 102 to prevent the scattered X-rays from directly entering the image sensor and damaging it.

[0024] An X-ray source intensity control unit 106 is connected to the X-ray source 101, which controls the intensity of the X-rays in response to commands from a calculation unit 130, which will be described later. The X-ray attenuation plate 102 is held on an insertion / removal stage 107, which can control its insertion / removal and position, and the sample holder 104 is also held on a similar sample stage 108, and both are connected to a stage control unit 109, which operates in response to commands from the calculation unit 130.

[0025] On the other hand, the image sensor 105, which is the main part of the camera unit, is connected to a camera control unit 120 having an exposure time control unit 121 and an image data readout unit 122, which exchange command signals and data with an arithmetic unit (computer including a control function) 130. The exposure time control unit 121 controls the exposure time so as to be able to deal with a wide range of conditions, from a condition in which at least one X-ray photon is incident on each pixel during 10,000 frames to a condition in which several tens of X-ray photons are discretely incident on the sensitive surface of the image sensor 105 during one frame. As an example, for an image sensor with 1024 x 1024 pixels, the average frequency of incident photons in each pixel is 10 -4 The image sensor 105 repeats imaging operations at a frame rate on the order of kHz, and the image data readout unit 122 converts the image signal from the image sensor 105 into a digital signal, reads it out at high speed, and sends it to the calculation unit 130.

[0026] When X-rays are irradiated from the X-ray source 101 onto the sample 103 held in the sample holder 104, the X-rays that pass through or are scattered by the sample 103 are incident on the sensitive surface of the image sensor 105 with an intensity that depends on the sample 103. The X-rays that are discretely incident on the sensitive surface of the image sensor 105 in photon units undergo photoelectric conversion in one or more pixels (unit light-receiving elements), and the converted electrical signals are digitized by a readout unit 122 built into or mounted in the image sensor 105 and read out as image data.

[0027] Note that an imaging optical system such as a pinhole may be inserted between the sample 103 and the image sensor 105 as appropriate depending on the resolution and imaging method required for the image to be captured.

[0028] (Charge sharing with image sensor) The image sensor 105 is configured as a pixel array type having a sensitive surface in which pixels (unit light receiving elements) that receive X-rays and perform photoelectric conversion are arranged two-dimensionally. The basic configuration of an image sensor for X-ray imaging is the same as that for visible light imaging, and can be configured as a semiconductor device with a CCD (Charge Coupled Device) or CMOS (Complementary Metal-Oxide-Semiconductor) configuration, for example. However, in the high-precision X-ray imaging system 100 that is the subject of the present invention, the number of X-ray photons in one shot irradiated in pulses from the XFEL as the X-ray source 101 is 10 10 ~10 11 As a result, the number of photons incident on the unit light-receiving elements that make up the sensitive surface of the image sensor 105 may reach multiple per frame depending on the exposure time. Therefore, in order to accurately grasp the energy of the incident X-ray photons, it is desirable for the image sensor 105 to be a charge-integration type image sensor that accumulates charge based on the radiation incident during the exposure time for each pixel and outputs a corresponding electrical signal, rather than a so-called radiation-counting type sensor.

[0029] The configuration of the image sensor 105 will be described in detail. As mentioned above, a plurality of pixels each having a photoelectric conversion function are two-dimensionally arranged as unit light receiving elements on the sensitive surface of the image sensor 105. FIG. 2 schematically shows a partial configuration of the sensitive surface 10 of the image sensor 105. For example, if the number of pixels on the sensitive surface 10 is 1 million, the pixels 11 are arranged two-dimensionally in the X and Y directions in 1024 rows and 1024 columns to accommodate digital processing. In this case, if the pixel area (area where pixels are arranged) on the sensitive surface has dimensions of 10 mm x 10 mm, the dimensions of each unit light receiving element (pixel) will be approximately 10 μm x 10 μm.

[0030] When one X-ray photon is incident on the sensitive surface 10 of the image sensor 105, that X-ray photon is incident on one of the pixels (each unit light-receiving element) 11. FIG. 3 schematically shows an example of pixel values ​​that appear in the pixels of the sensitive surface 10 when an X-ray photon is incident. For simplicity's sake, we will focus on an area 12 shown in the upper diagram of FIG. 3, which includes 5 × 5 = 25 pixels, and assume that the X-ray photon is incident on pixel 11A, which is shown with diagonal lines to the left, in the center of area 12. When pixel 11A receives the X-ray photon, a portion of the signal resulting from the reception of the X-ray photon also appears in nearby pixel 11B, which is shown with diagonal lines to the right, due to a charge sharing phenomenon specific to semiconductor devices. The area occupied by pixels 11A and 11B where photoelectric conversion signals derived from incident X-ray photons appear varies depending on the X-ray intensity, etc., but it is expected that it may be, for example, 3 × 3 = 9 pixels, 5 × 5 = 25 pixels, or 7 × 7 = 49 pixels with pixel 11A at the center. What is important is that the signal components that leak into peripheral pixel 11B of pixel 11A at the center of the X-ray incidence also contain significant information for X-ray image analysis.

[0031] The lower part of Fig. 3 shows an example of the distribution of pixel signal values ​​included in region 12. As shown, pixels with large signal values ​​including signals derived from incident photons are distributed around pixel 11A where X-rays are incident, and pixels with erroneous signal values ​​output due to noise are distributed around the outer edge.

[0032] As described above, the image sensor 105 may be provided with a cooling mechanism such as a Peltier element to maintain a constant operating temperature and reduce noise. Alternatively, the image sensor 105 may be provided with a cooling mechanism in a configuration in which the image sensor 105 is housed in a vacuum airtight container with a light receiving window.

[0033] Now, let us consider X-rays detected by the image sensor 105. As described above, in order for charge sharing to occur when one X-ray photon is incident, two or more electron-hole pairs must be generated in response to the incidence of one photon in the semiconductor used in the photoelectric conversion element that constitutes the pixel.

[0034] For example, according to Non-Patent Document 9, the electron-hole pair creation energy E iis empirically expressed by the following formula:

number

[0035] In view of the above, in order to reliably generate charge sharing in the semiconductor material used for detecting X-rays, it is preferable that the energy of X-rays is 10 eV or more. Note that the photons incident on the image sensor 105 are not limited to X-rays, and may be electromagnetic radiation other than X-rays having a radiation energy of 10 eV or more, or particle radiation (charged particle beams and uncharged particle beams) having a kinetic energy of 10 eV or more.

[0036] (Image signal processing system and its operation) Next, the image signal processing system which plays a central role in the present invention and its operation will be described. In the X-ray imaging system 100 of FIG. 1, a group of digitized data signals from the image data readout unit 122 is sent to an image signal processing unit 140 of a calculation unit (computer) 130 equipped with information processing functions. The image signal processing unit 140 is configured to perform a predetermined determination process (described later) on the detection signal acquired from the image sensor 105 to acquire an image by X-ray imaging. That is, the image signal processing unit 140 performs a noise component removal process based on the present invention, and a group of high-quality image data containing only significant signals is supplied to a data compression unit 132. As a result, efficient data compression is performed on a large amount of image data. The compressed group of image data is stored in a data storage unit 133 for analysis and, if necessary, transmitted to an analysis center or the like. Meanwhile, the processed image signals can be displayed in real time as an X-ray image on an image display unit 150 for monitoring.

[0037] Here, the image signal processing unit 140 receives a detection signal from a pixel 11A on the light receiving surface of the image sensor 105 where an X-ray photon is incident, or from a pixel 11B adjacent to the pixel where the detection signal is generated by charge sharing. As described above, the received data contains many pixels that exhibit signal values ​​other than 0 due to noise inherent in semiconductor devices. If such pixel values ​​due to noise are left, the data size will increase. The image signal processing unit 140 according to this embodiment separates pixels containing significant information derived from incident photons from pixels containing only pixel values ​​due to noise, corrects the significant information lost by the separation as necessary, removes pixel values ​​due to noise, and outputs data suitable for data compression.

[0038] 4 shows a schematic functional configuration of the image signal processing unit 140. The image signal processing unit 140 includes an input data storage unit 141, a pixel selection unit 142, a threshold discrimination unit 143, a constant setting unit 144, a data setting unit 145, a data storage unit 146, and an image data transmission unit 147.

[0039] The input data storage unit 141 is configured as, for example, a frame memory and temporarily stores image data read from the image sensor 105 (input data IN in FIG. 4). The pixel selection unit 142 appropriately selects pixels to be processed from the image data stored in the input data storage unit 141. As will be described later, the threshold discrimination unit 143 performs threshold discrimination processing on the selected pixels to distinguish between significant signals and noise using constants such as thresholds read from a constant setting unit 144 (also referred to as a threshold setting unit). The constant setting unit 144 includes a statistical processing unit 148 that statistically calculates a threshold corresponding to an observed event based on previously acquired experimental data, etc., and a constant storage unit 149 that stores the calculated constants. Note that the statistical processing unit 148 that calculates the thresholds, a feature of the present invention, may be configured as an independent computer. The data setting unit 145 sets output data corresponding to each pixel based on the results of the threshold discrimination processing. The output data is temporarily stored in the data storage unit 146. When the image data sending unit 147 has completed processing of, for example, one frame of image data, it reads out output data corresponding to the pixels of one frame from the data storage unit 146 and sends it to the data compression unit 132 of the calculation unit 130 and the image display unit 150.

[0040] (Noise removal process according to the first embodiment: 2×2-sum method) The noise removal process in the image signal processing unit 140 according to the embodiment will be described below. The noise removal process technique according to the first embodiment is also referred to as the 2×2-sum method. In the 2×2-sum method, one block including M×N pixels (M vertical × N horizontal) is processed as the pixels of the processing target area, and discrimination is applied to the sum of the pixel values ​​of the block area using a 2×2-sum threshold (T1). If the result is true, all pixel values ​​in the block are retained as significant information. On the other hand, if all discrimination results are false, the pixel values ​​of all pixels are discriminated as noise and excluded from the data.

[0041] In the following, an example will be described in which one block includes four pixels, 2 vertically and 2 horizontally. Fig. 5 shows a flowchart of the noise removal process by the image signal processing unit 140 in the first embodiment. In the following, in an area on the sensible surface where pixels to be subjected to noise removal process by the image signal processing unit 140 are arranged (hereinafter referred to as the processing target area), the number of pixels in the row direction (Y direction) is H, the number of pixels in the column direction (X direction) is W, the row number is i, and the column number is j. The pixel value of the input data of the pixel in the i-th row and j-th column of the processing target area is I(i,j), and the pixel value of the output data is O(i,j).

[0042] Step SA1 First, the threshold discriminator 143 reads out a 2×2-sum threshold T1 for noise removal processing selected in accordance with the corresponding observation event from the constant memory 149 of the constant setting unit 144, and sets the 2×2-sum threshold T1 in the threshold discriminator 143 as the threshold to be used for threshold discrimination. The 2×2-sum threshold T1 is a value statistically calculated in advance by the statistical processing unit 148 included in the constant setting unit 144 based on various conditions such as the physical properties of the observation sample, the intensity of the irradiated X-rays, and the frame period, and is not uniquely determined. That is, the statistically determined 2×2-sum threshold T1 used in the present invention is characterized in that, unlike a threshold uniquely determined based on the noise level characteristics specific to the image sensor, the statistical processing unit 148 calculates a large amount of data previously acquired under various conditions such as the physical properties of the observation sample, the intensity of the irradiated X-rays, and the frame period in advance, and dynamically sets the selected value in the threshold discriminator 143 for each individual observation event from among multiple constants stored in the constant memory 149.

[0043] Step SA2 The data setting unit 145 resets the pixel value of the output data for each pixel in the processing target area, i.e., sets O(i, j) = 0. The output data for each pixel is stored in the data storage unit 146, and the data setting unit 145 can update the output data as needed.

[0044] Step SA3 The pixel selection unit 142 sets the row number i and column number j for specifying pixels in the processing target area to 1 as initial values ​​(i=1, j=1).

[0045] Step SA4 The pixel selection unit 142 acquires the pixel value of the i-th row and j-th column in the processing target area from the input data storage unit 141, and inputs it to the threshold discrimination unit 143 as input data I(i,j).

[0046] Step SA5 The threshold discrimination unit 143 determines whether the sum of pixel values ​​of a block including the pixel in the i-th row and j-th column of the processing target area and adjacent pixels is equal to or greater than the 2×2-sum threshold T1. Note that here, the determination is made for the sum of pixels included in the area from the i-th row to the (i+M-1)-th row and the j-th row to the (j+N-1)-th column. In this case, the threshold discrimination unit 143 determines whether the following formula is satisfied:

number

number

[0047] Step SA6 If the determination result is false, the data setting unit 145 determines the values ​​of the input data of the pixels included in the determined block as the output data of those pixels, as shown in the following equations.

number

number

number

number

number

[0048] Step SA7 The pixel selection unit 142 determines whether the column number j is smaller than "W-N+1".

[0049] Step SA8 If the column number j is smaller than "W-N+1", the pixel selection unit 142 adds "1" to the column number j (j=j+1) and returns the process to step SA4.

[0050] Step SA9 If the column number j is the same as "W-N+1", the pixel selection unit 142 determines whether the row number i is smaller than "H-M+1".

[0051] Step SA10 If the row number i is smaller than "H-M+1", the pixel selection unit 142 adds "1" to the row number i (i=i+1), sets the column number j to "1", and returns the process to step SA4. If the row number i is the same as "H-M+1", the threshold discrimination unit 143 determines that the noise removal process for one frame (one processing target region) of image data has been completed, and proceeds to step SA11.

[0052] Step SA11 If it is determined in step SA10 that the row number i is the same as "H-1," that is, if it is determined that the noise removal process is completed, the image data sending unit 147 reads the output data O(i,j) after the noise removal process is completed from the data storage unit 146. Then, the image data sending unit 147 reads the output data of the pixels in the processing target area and outputs it to the data compression unit 132 and the image display unit 150. The output data is written to a storage device (for example, the data saving unit 133 or a storage device not shown) as appropriate, and can also be displayed on the image display unit 150.

[0053] By repeating the process consisting of steps SA4 to SA10, the processing target area (area 12 in FIG. 2) is swept using a block of four pixels, consisting of the selected pixel and its three adjacent pixels. FIG. 6 shows the order of sweeping the processing target area using blocks of 2×2=4 pixels in the noise removal process according to the first embodiment. By sequentially sweeping the processing target area on the sensible surface using blocks of four adjacent pixels in this way, it is possible to process the data of all pixels.

[0054] (Comparison between the 2x2-sum method and conventional methods) Next, the effect of noise processing according to this embodiment will be explained in comparison with a general conventional method. First, an example of noise removal processing using general threshold discrimination will be explained. FIG. 7 shows the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal processing using general threshold discrimination. Here, the energy amount of one incident X-ray photon is normalized to "1," and signals of intensity corresponding to the detected energy appear in the incident unit photoreceptor element and surrounding unit photoreceptor elements that cause charge sharing. In other words, if the influence of noise is not taken into consideration, the sum of the signal intensities of unit photoreceptor elements that appear in response to the incidence of an X-ray photon will ideally be equal to "1," which is the normalized energy amount of one X-ray photon.

[0055] In this example, each pixel is selected in turn, the sum of the pixel values ​​of the selected pixel and its adjacent pixels (up to eight) is calculated, and the calculated sum is compared with a threshold. If the calculated sum is smaller than the threshold, the pixel value of the selected pixel is set to 0, and if the calculated sum is equal to or greater than the threshold, the pixel value of the selected pixel is set to the calculated sum. In this case, when the central pixel in the third row and third column is selected, the sum (1.01) is greater than the threshold, so the pixel value of this pixel is set to 1.01 and the pixel values ​​of the other pixels are set to 0.

[0056] This eliminates the effects of noise while preserving the signal strength at 1.01, which is close to the value of 1 that should be detected when an X-ray photon strikes the object. However, this method also results in a significant loss of information about the signal distribution that appears when an X-ray photon strikes the object, significantly reducing the spatial resolution of the captured image.

[0057] Next, we will explain the case where the 2x2-sum method is applied to the same noise distribution as the conventional method. Figure 8 shows the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal processing using the 2x2-sum method. ph :However, Q ph is the number of electrons generated by one radiation, which corresponds to the signal level as a pixel value. ph =1]. In this example, the pixel values ​​of the pixel in the third row and third column (pixel value 0.73) and the eight pixels adjacent to it are preserved as is. The sum of the pixel values ​​of the nine pixels whose values ​​are preserved is 1.01.

[0058] A simulation of small-angle X-ray scattering was performed for the noise removal according to this embodiment and the general noise removal described above. In the simulation, the photon energy of the X-ray was set to 10 KeV, and the average frequency of incident photons in the image was set to 10 -4 The average value of 100 images was calculated using photons / pixel / frame. In general noise reduction, the loss of photon signals is -1.3 × 10 -3The data compression rate was 6000. The loss rate of the photon signal here is the value obtained by subtracting 1 from the ratio of detected photons to the total number of incident photons. On the other hand, in the 2×2-sum method according to this embodiment, the loss rate of the photon signal is −1.3×10 -3 It was confirmed that the data compression rate could be about 1000 times higher than without noise removal processing.

[0059] From the above, it was verified that this configuration can significantly suppress the loss of photon signals due to noise removal, while compressing data to a level roughly equivalent to that of general data.

[0060] Therefore, with this configuration, compared to a general method that performs threshold discrimination on the sum of pixel values ​​of pixels included in a certain region, the sum of signal intensities is preserved, i.e., incident energy information can be accurately preserved. Furthermore, since the pixel values ​​of pixels with large pixel values ​​and their neighboring pixels can be preserved, the signal intensity distribution can also be suitably preserved. Therefore, compared to general conventional methods, the loss of significant information that accompanies noise removal signal processing can be effectively reduced, enabling compression processing with high signal accuracy.

[0061] Embodiment 2 The following describes an X-ray image signal processing method according to embodiment 2. The X-ray imaging system according to this embodiment has a configuration generally similar to that of embodiment 1, but as will be described below, two thresholds read from a constant setting unit 144 are set in a threshold discriminator 143, and a correction constant C is set in a data setting unit 145.

[0062] (Noise removal processing according to the second embodiment: T2C method) The noise removal process according to this embodiment is called the T2C (Threshold to Correction) method, in which threshold discrimination is applied to each pixel in two stages, and the amount of charge loss estimated in the first stage of threshold discrimination (first T2C threshold T2) is distributed and added in the second stage of threshold discrimination (second T2C threshold T3).

[0063] 9 shows a flowchart of the noise removal process according to the second embodiment. In the following, as in the first embodiment, the number of pixels in the row direction (Y direction) of the processing target area on which the image signal processing unit 240 performs the noise removal process is defined as H, the number of pixels in the column direction (X direction) as W, the row number as i, and the column number as j. The pixel value of the input data of the pixel in the i-th row and j-th column of the processing target area is defined as I(i,j), and the pixel value of the output data as O(i,j).

[0064] Step SB1 As an initial setting, the threshold discriminator 143 reads a first T2C threshold T2 and a second T2C threshold T3 for noise removal processing from the constant memory unit 149 of the constant setting unit 144, and sets them in the threshold discriminator 143 as thresholds to be used in the threshold discrimination processing. Similarly, the data setting unit 145 reads a correction constant C from the constant memory unit 149 of the constant setting unit 144. Unlike thresholds that are uniquely determined based on noise level characteristics specific to the image sensor, such as the 2×2sum threshold T1 in the first embodiment, the first T2C threshold T2, the second T2C threshold T3, and the correction constant C are obtained by statistically processing a large amount of data acquired in advance under various conditions such as the physical properties of the observation sample, the intensity of irradiated X-rays, and the frame period in the statistical processing unit 148, and are not uniquely determined. Here, as an example, the first T2C threshold T2 is set to [T2=14σ: where σ is the noise level specific to the image sensor], and the second T2C threshold T3 is set to [T3=7σ], which is lower than the first T2C threshold T2.

[0065] Step SB2 The data setting unit 145 resets the pixel value of the output data for each pixel in the processing target region, that is, sets O(i,j)=0.

[0066] Step SB3 The pixel selection unit 142 sets the row number i and column number j for specifying pixels in the processing target region to 1 as initial values ​​(i=1, j=1).

[0067] Step SB4 The pixel selection unit 142 acquires the input data I(i, j), which is the pixel value of the pixel at the i-th row and j-th column, from the input data storage unit 141 and inputs it to the threshold discrimination unit 143.

[0068] Step SB5 The threshold discrimination unit 143 determines whether the pixel value I(i, j) of the pixel at the i-th row and j-th column is greater than or equal to the first T2C threshold T2 (I(i, j) ≥ T2).

[0069] Step SB6 If it is determined in step SB5 that the pixel value I(i, j) is greater than or equal to the first T2C threshold T2 (I(i, j) ≥ T2), the data setting unit 145 determines the pixel value I(i, j) of the pixel at the i-th row and j-th column as the output data O(i, j) corresponding to the pixel at the i-th row and j-th column as it is, as shown in the following formula.

Equation

[0070] Step SB7 If it is determined in step SB5 that the pixel value I(i, j) is less than the first T2C threshold T2 (I(i, j) < T2), the threshold discrimination unit 143 discriminates whether the pixel value I(i, j) is greater than or equal to the second T2C threshold T3 (I(i, j) ≥ T3). If it is determined that the pixel value I(i, j) is less than the second T2C threshold T3 (I(i, j) < T3), the data setting unit 145 determines that the pixel value I(i, j) of the pixel at the i-th row and j-th column is affected by noise, maintains the output data O(i, j) corresponding to the pixel at the i-th row and j-th column as 0, and proceeds with the process in step SB10.

[0071] Step SB8 If it is determined in step SB7 that the pixel value I(i, j) is greater than or equal to the second T2C threshold T3 (I(i, j) ≥ T2), the data setting unit 145 assigns the correction constant C to the output data O(i, j) corresponding to the pixel at the i-th row and j-th column (O(i, j) = C).

[0072] Step SB9 After steps SB6 and SB8, the threshold discrimination unit 143 determines whether the column number j of the pixel at the i-th row and j-th column is smaller than "W" (j < W).

[0073] Step SB10 If the column number j of the pixel at the i-th row and j-th column is smaller than "W", the pixel selection unit 142 adds 1 to the column number j (j = j + 1) and returns the process to step SB4.

[0074] Step SB11 If it is determined in step SB9 that the column number j of the pixel at the i-th row and j-th column is the same as "W" (j = W), the pixel selection unit 142 determines whether the row number i of the pixel at the i-th row and j-th column is smaller than "H" (i < H).

[0075] Step SB12 If it is determined in step SB11 that the row number i of the pixel at the i-th row and j-th column is smaller than "H", the pixel selection unit 142 adds 1 to the row number i (i = i + 1) and returns the process to step SB4.

[0076] Step SB13 If it is determined in step SB12 that the row number i of the pixel at the i-th row and j-th column is the same as "H", the image data transmission unit 147 reads the output data O(i, j) after the noise removal process from the data storage unit 146. Then, the image data transmission unit 147 reads the output data of the pixels in the processing target area and outputs it to the data compression unit 132 and the image display unit 150. The output data is appropriately written into a storage device (for example, the data storage unit 133 or a storage device not shown) and can also be displayed on the image display unit 150.

[0077] By performing the repeated process composed of the above steps SB4 to SB12, the selected pixels will sweep the area 12. FIG. 10 shows the order of pixel sweeping in the noise removal process according to the second embodiment. According to the order shown in FIG. 10, all the pixels in the area 12 can be swept.

[0078] Next, a method for determining the correction constant C will be described. The image signal processing unit 140 may determine the correction constant C in advance by the method described below. By the noise removal process described with reference to FIG. 9, the pixel value I(i,j) of the i-th row and j-th column, which is input data, is converted into output data O(i,j) as follows:

number

[0079] The correction constant C may be determined so that the sum of the signal intensities contained in the image after noise processing is equal to the sum of the signals corresponding to the incident X-rays. In this case, the correction constant C is expressed by the following formula:

number

number

[0080] Q true However, if we assume that the noise detected by the image sensor 105 is zero on average, it can be estimated using the pixel values ​​of the image before noise removal processing. For example, as expressed by the following equation, the sum of the pixel values ​​of the image before noise removal processing is Q true may be used as an estimate of

number

[0081] When the noise in the image sensor 105 varies greatly, it is expected that the estimation using the above formula will result in a large error. In this case, for example, as expressed by the following formula, the sum of the pixel value saved after the noise removal processing according to the second embodiment and the pixel values ​​around it is set to Q true It may also be an estimate of

number

[0082] (Comparison of T2C method and conventional method) Next, the effect of noise processing according to this embodiment will be explained in comparison with a general method. First, an example of noise processing using general threshold discrimination will be explained. FIG. 11 shows the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal using general threshold discrimination. The distribution of detection signals and normalization of signal intensity are the same as in the first embodiment.

[0083] In this example, pixels with a signal strength of 0.14 or less are considered to be noise and their pixel values ​​are set to 0. In this case, only the pixel values ​​(0.16) of the pixels in the third row, third column (strength 0.73) and the fourth row, third column are saved, and the values ​​of all other pixels are corrected to 0.

[0084] This method can remove the effects of noise, but the sum of the signal intensities after noise removal is 0.73 + 0.16 = 0.89, which is 0.11 smaller than the value "1" that should be detected when an X-ray photon is incident, resulting in a significant loss of information about the incident energy. This leads to variations in the sum of the remaining signal intensities for X-ray photons incident at different positions. Furthermore, a significant loss of information about the signal distribution that appears when an X-ray photon is incident results in a significant reduction in the spatial resolution of the captured image.

[0085] 12 shows the distribution of pixel values ​​before noise removal and the distribution of pixel values ​​after noise removal processing using the T2C method. In this embodiment, the pixel value of the pixel in the third row, third column (pixel value 0.73) is equal to or greater than the first threshold value T2, so the pixel value is stored as is. Furthermore, the pixel value of the pixel in the fourth row, third column (pixel value 0.16) is smaller than the first threshold value T2 and larger than the second threshold value 0.15, so in step SB5 the pixel value is corrected by the correction constant C to 0.27. Therefore, the sum of the pixel values ​​of the third row, third column and the fourth row, third column is 1.

[0086] Simulations of small-angle X-ray scattering were performed for noise removal according to the second embodiment and general noise removal. The simulations were the same as those in the first embodiment, so a description of the conditions will be omitted. In the general noise removal, the degree of loss of the photon signal was 6.0 × 10 -2 , the data compression rate is 3.3 × 10 3 On the other hand, in the noise removal (T2C method) according to the second embodiment, the degree of loss of the photon signal was 0.5×10 -3 The data compression rate was significantly reduced to , and a value comparable to that of the conventional method was obtained.

[0087] From the above, it has been verified that this configuration can significantly reduce the loss of photon signals due to noise removal, while compressing data more effectively than conventional methods.

[0088] This allows the sum of signal intensities to be preserved, that is, the information on incident energy to be preserved accurately, compared to when general threshold discrimination is performed for each pixel.

[0089] Embodiment 3 The noise removal process according to the third embodiment will be described. The noise removal process according to the third embodiment is a combination of the above-mentioned 2×2-sum and T2C methods. Here, the noise removal process is also referred to as an integrated method.

[0090] The X-ray imaging system of this embodiment has a configuration generally similar to that of embodiment 1, but as described below, it is configured such that two thresholds are set for the threshold discrimination unit 143 by the constant setting unit 144, and a correction constant C is set for the data setting unit 145. (Noise Removal Processing According to the Third Embodiment: Integration Method)

[0091] An X-ray image signal processing method according to the third embodiment will be described below. FIG. 13 shows a flowchart of the noise removal process (integration method) according to the third embodiment. In the integration method, threshold discrimination is performed using 2×2-sum, and if the result of the threshold discrimination is false, threshold discrimination using the T2C method is further performed. In other words, the integration method can be considered as performing the first-stage threshold discrimination and data setting in the T2C method in the same way as the 2×2-sum method, and then performing the second-stage threshold discrimination for each pixel in the block using the 2×2-sum method if the result of the first-stage threshold discrimination is false. In other words, by combining threshold discrimination for each pixel and threshold discrimination for the sum of pixel values ​​within a predetermined region, it is possible to remove noise components while maintaining the accuracy of the image signal, even when pixel signals with significantly different intensities are mixed.

[0092] Step S1 As an initial setting, the threshold discriminator 143 receives a 2×2sum threshold T A and T2C threshold T B The data setting unit 145 reads the correction constant C from the constant setting unit 144 and sets it as the threshold value used in the threshold discrimination process. A corresponds to the 2×2sum threshold T1 according to the first embodiment, and the T2C threshold T B corresponds to the second T2C threshold T3 according to the second embodiment.

[0093] Steps S2 to S4 Steps S2 to S4 are the same as steps SA2 to SA4 (FIG. 5) according to the first embodiment, respectively, and therefore will not be described.

[0094] Step S5 The threshold discrimination unit 143 determines whether the sum of pixel values ​​of a block including the pixel at row i and column j in the processing target area and its adjacent pixels is equal to or exceeds a 2×2-sum threshold T A It is to be noted that, similarly to step SA5 (FIG. 5) according to the first embodiment, the determination is made for the sum of pixels included in the area from the i-th row to the (i+M-1)-th row and from the j-th row to the (j+N-1)-th column.

[0095] Step S6 Step S6 is the same as step SA6 (FIG. 5) according to the first embodiment, and therefore a description thereof will be omitted.

[0096] Step S7 If the determination result in step S5 is false, that is, if the sum of the pixel values ​​in the region including the pixel in the i-th row and j-th column and the adjacent pixels is less than the 2×2sum threshold T A If m is smaller than m, the pixel selection unit 142 sets the row number m in the block to 1 and the column number n in the block to 1 (m=1, n=1).

[0097] Step S8 The pixel selection unit 142 acquires the pixel value of the (i+m−1)th row and the (j+n−1)th column from the input data storage unit 141 .

[0098] Step S9 The threshold discriminator 143 determines whether the pixel value in the (i+m-1)th row and the (j+n-1)th column is a T2C threshold T B Is it greater than or equal to (I(i+m-1,j+n-1)≧T B ) is determined.

[0099] Step S10 The pixel value I is the T2C threshold T B More than (i+m-1,j+n-1)≧T B ), the data setting unit 145 assigns a correction constant C to the output data O(i+m-1, j+n-1) (O(i+m-1, j+n-1)=C).

[0100] Step S11 The pixel selection unit 142 determines whether the column number n within the block is smaller than "N".

[0101] Step S12 If the intra-block column number n is smaller than "N", the pixel selection unit 142 adds "1" to the intra-block column number n (n=n+1) and returns the process to step S8.

[0102] Step S13 If the intra-block column number n is the same as "N", the pixel selection unit 142 determines whether the intra-block row number m is smaller than "M". If the intra-block row number m is the same as "M", the pixel selection unit 142 proceeds to step S15.

[0103] Step S14 If the intra-block row number m is smaller than "M", the pixel selection unit 142 adds "1" to the intra-block row number m (m=m+1) and sets the intra-block column number n to "1", and returns the process to step S8.

[0104] Steps S15 to S19 Steps S15 to S16 are the same as steps SA7 to SA11 (FIG. 5) according to the first embodiment, respectively, except that the process returns to step S4 after steps S16 and S18, and therefore a description thereof will be omitted.

[0105] By repeating the process consisting of steps S4 to S17, the sensitive surface 10 is swept by an M×N pixel region including the selected pixel. FIG. 6 shows the order of sweeping the target region of the sensitive surface when M×N pixels are one block in the noise removal process according to the first embodiment. For simplicity, M=N=2 is set in FIG. 7. In this way, by sequentially sweeping the sensitive surface for a unit region consisting of M×N pixels, it is possible to process the data of all pixels.

[0106] As described above, according to this configuration, by applying T2C to each block after performing threshold discrimination for each block using the 2x2-sum algorithm, data setting for each pixel in the block can be performed more precisely than in the first embodiment. This makes it possible to remove noise while preserving the signal distribution in the input data.

[0107] Other embodiments The present invention is not limited to the above-described embodiments and may be modified as appropriate without departing from the spirit and scope of the present invention. For example, while the configuration of each component of the X-ray imaging system, including the image signal processing unit, has been described primarily as a hardware configuration, the present invention is not limited to this. Processing in each component of the X-ray imaging system can also be realized by having a CPU (Central Processing Unit) execute a computer program. The program includes instructions (or software code) that, when loaded into a computer, cause the computer to execute one or more functions described in the above-described embodiments. The program may be implemented and executed on an electrical circuit within the detector. For example, the program may be implemented in an integrated circuit within the sensor, a field-programmable gate array (FPGA) within the detector, or an application-specific integrated circuit (ASIC). The program may also be stored on a non-transitory computer-readable medium or a tangible storage medium, or provided as a software package or a downloadable program product.

[0108] The above-mentioned computer can be realized by a computer such as a dedicated computer or a personal computer (PC). Fig. 14 shows an example of the configuration of a computer 1000. However, the computer does not need to be physically a single computer, and may be multiple computers when performing distributed processing. As shown in Fig. 14, the computer 1000 has a CPU (Central Processing Unit) 1001, a ROM (Read Only Memory) 1002, and a RAM (Random Access Memory) 1003, which are interconnected via a bus 1004. Note that although explanation of OS software for operating the computer will be omitted, it is assumed that the computer that constitutes this information processing device also has such software.

[0109] An input / output interface 1005 is also connected to the bus 1004. To the input / output interface 1005, for example, an input unit 1006 including a keyboard, mouse, sensor, etc., a display including a CRT, LCD, etc., an output unit 1007 including headphones, speakers, etc., a storage unit 1008 including a hard disk, etc., and a communication unit 1009 including a modem, terminal adapter, etc. are connected.

[0110] The CPU 1001 executes various processes in accordance with various programs stored in the ROM 1002 or various programs loaded from the storage unit 1008 to the RAM 1003. In this embodiment, for example, it executes the processes of each unit of the X-ray imaging system described below. A GPU (Graphics Processing Unit), FPGA, or ASIC may be provided separately from the CPU 1001 and be made to perform the same processes as the CPU 1001. Note that the GPU is suitable for performing routine processes in parallel, and by applying it to the learning process described below, it is possible to improve the processing speed compared to the CPU 1001. Furthermore, by implementing the processes in the FPGA and ASIC, it is possible to perform the processes faster and with less power consumption compared to the CPU 1001. The RAM 1003 also stores data necessary for the CPU 1001 and GPU to execute various processes.

[0111] The communication unit 1009 performs communication processing via the Internet (not shown), for example, transmits data provided by the CPU 1001, and outputs data received from a communication partner to the CPU 1001, RAM 1003, and storage unit 1008. The storage unit 1008 exchanges data with the CPU 1001 and stores and erases information. The communication unit 1009 also performs communication processing of analog or digital signals with other devices.

[0112] The input / output interface 1005 is also connected to a drive 1010 as needed, and, for example, a magnetic disk 1011, an optical disk 1012, a flexible disk 1013, or a semiconductor memory 1014 is appropriately attached, and computer programs read from these are installed in the memory unit 1008 as needed.

[0113] Although the above embodiment has been described as a representative example of an X-ray imaging system, the present invention is widely applicable to other radiation imaging systems that use semiconductor image sensors. Incidentally, efficient data compression for data storage is an important process even in compact X-ray imaging systems that use low-output X-ray tubes as radiation sources, so the noise reduction process of the present invention is extremely useful. Furthermore, the present invention is applicable not only to imaging devices that use a direct X-ray irradiation method, but also to imaging devices that use an indirect X-ray method, in which an X-ray image is converted into a visible light image using a scintillator and then captured.

[0114] The imaging method in the X-ray imaging system according to the above-described embodiment can also be applied to an imaging method of direct imaging using electron beam irradiation, such as an electron microscope, or a case where neutron beams are captured by a semiconductor image sensor coated with a thin film containing boron as the main element.

[0115] Furthermore, in the above-described embodiment, the pixel value of a single pixel or the sum of pixel values ​​within a 2×2 pixel region is the target of threshold discrimination, but the target pixel region is not limited to these. For example, the sum of pixel values ​​within an M×N pixel region or the sum of pixel values ​​within a pixel region along a pre-estimated trajectory of radiation incident on the detector may also be the target of threshold discrimination. Such threshold discrimination is particularly useful in detecting, for example, electron beams, neutron beams, charged particle beams, and high-energy X-rays.

[0116] The 2x2-sum method described in the first embodiment is suitable for image processing in radiation observations that use a relatively high-energy radiation source (SACLA: SPring-8 Angstrom Compact Free Electron Laser) at the synchrotron radiation facility where the inventor works. On the other hand, the T2C method described in the second embodiment is suitable for image processing in radiation observations that use a relatively low-energy radiation source. The integration method described in the third embodiment is suitable for use as software when configured as a general-purpose computer program.

[0117] In the above embodiment, the comparison and discrimination between a threshold value and a pixel value has been described, but when the threshold value and the pixel value are identical, the pixel value may be treated as either exceeding the threshold value or as being below the threshold value. In other words, when determining which of two values ​​is larger or smaller and obtaining two determination results, if the two values ​​are equal, the value may be included in either of the two determination results as necessary.

[0118] This application claims priority based on Japanese Patent Application No. 2021-175897, filed on October 27, 2021, the disclosure of which is incorporated herein in its entirety. [Explanation of symbols]

[0119] 10 Sensitive surface 11, 11A, 11B pixels 12 areas 100 X-ray imaging system 101 X-ray source 102 X-ray attenuation plate 103 samples 104 Sample holder 105 Image Sensor 106 X-ray intensity control unit 107 Insertion / Extraction Stage 108 Sample Stage 109 Stage control unit 120 Camera control unit 121 Exposure time control section 122 Image data reading unit 130 Arithmetic section 132 Data Compression Unit 133 Data Storage Unit 140 Image signal processing unit 141 Input data storage unit 142 Pixel selection unit 143 Threshold discrimination unit 144 Constant setting section 145 Data setting section 146 Data storage unit 147 Image data transmission unit 148 Statistical Processing Unit 149 Constant storage section 150 Image display unit 1000 computers 1001 CPU(Central Processing Unit) 1002 ROM (Read Only Memory) 1003 RAM (Random Access Memory) 1004 Bus 1005 Input / Output Interface 1006 Input section 1007 Output section 1008 Storage section 1009 Communications Department 1010 Drive 1011 Magnetic Disk 1012 Optical disc 1013 Flexible Disk 1014 Semiconductor Memory

Claims

1. A radiation imaging system comprising an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and which outputs signals electrically converted by the pixels in response to incidence of radiation containing image information as pixel values, A method for separating one or more pixels that output a significant electrical conversion signal derived from incident radiation from pixels that output a signal due to noise inherent in a sensor device, based on a predetermined threshold value, comprising: Adding pixel values ​​of adjacent blocks of pixels in the image sensor, each block consisting of a plurality of pixels in both the vertical and horizontal directions, to obtain a total sum; The sum is sequentially compared with a predetermined threshold value for the sum of pixel values ​​in a block region, and pixel values ​​of each pixel in a block whose sum is lower than the threshold value are removed as pixel data caused by noise inherent to the device, and pixel values ​​of each pixel included in a block whose sum is higher than the threshold value are output as significant pixel signal data. Radiation image signal processing method.

2. A sum of pixel values ​​is obtained for each block, which is four pixels, two pixels vertically and two pixels horizontally, and if the sum is equal to or greater than a threshold value for the sum of pixel values ​​in the block region, the pixel values ​​of the four pixels in the block are maintained as significant signals; The pixel values ​​of the pixels of the blocks whose sum is smaller than a threshold value for the sum of pixel values ​​of the block region in all block determinations are determined to be noise and are removed.

2. The radiation image signal processing method according to claim 1.

3. The threshold value for the pixel value of the block region is dynamically set as a threshold value for each individual event for which radiographic image observation is performed.

3. The radiation image signal processing method according to claim 1.

4. A radiation imaging system comprising an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, the image sensor converting signals containing image information into electricity by the pixels in response to incidence of the radiation, and outputting the converted signals as pixel values, A method for separating one or more pixels that output a significant electrical conversion signal derived from incident radiation from pixels that output a pixel value due to noise inherent in a sensor device, based on a predetermined threshold value, comprising: an output signal for each pixel from the image sensor is compared with a predetermined first threshold value as a noise level, and image signals of pixels having pixel values ​​equal to or greater than the first threshold value are obtained; comparing pixel values ​​of pixels having pixel values ​​smaller than the first threshold value with a second threshold value that is predetermined as a noise level lower than the first threshold value; generating an image signal by adding a correction value determined by arithmetic processing of a plurality of previously acquired data to the pixel values ​​of pixels having pixel values ​​smaller than the first threshold value and equal to or greater than the second threshold value; an image signal of the pixel value of the pixel to which the correction value has been added is output together with an image signal of a pixel having a pixel value equal to or greater than the first threshold value; Radiation image signal processing method.

5. the first and second threshold values ​​and the correction value are dynamically set for each individual event in which a radiographic image is observed, pixel values ​​due to noise are removed, and pixel values ​​of significant signals are output as radiographic image data.

5. The radiation image signal processing method according to claim 4.

6. A radiation imaging apparatus includes an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and outputs, as pixel values, signals electrically converted by the pixels in response to incident radiation as image information, a constant setting unit that determines a threshold value for detecting noise components appearing in each pixel; a pixel that outputs a significant electrical conversion signal derived from incident radiation as a pixel value based on the threshold value from the constant setting unit; a threshold discriminator that discriminates pixels that output pixel values ​​due to noise inherent to the sensor device; the constant setting unit includes a statistical processing unit that statistically processes a plurality of data acquired in advance to calculate a plurality of constants, and a constant storage unit that stores the calculated plurality of constants, and a constant selected for each observation event is set as a threshold in the threshold discrimination unit. Radiation image signal processing device.

7. a radiation source that irradiates the observation sample with radiation; a camera unit including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged; an image signal processing unit that outputs, as pixel values, signals electrically converted by the pixels in response to incidence of radiation containing information about the observation sample; The image signal processing unit a constant setting unit that determines a threshold value for detecting noise components appearing in each pixel; a pixel that outputs a significant electrical conversion signal derived from radiation as a pixel value based on the threshold value from the constant setting unit; a threshold discriminator that discriminates pixels that output pixel values ​​due to noise inherent to the sensor device; the constant setting unit includes a statistical processing unit that statistically processes a plurality of data acquired in advance to calculate a plurality of constants, and a constant storage unit that stores the calculated plurality of constants, and a constant selected for each observation event is set as a threshold in the threshold discrimination unit. Radiography imaging system.

8. the radiation source is an X-ray source that irradiates X-rays, The image signal processing unit a camera control unit for acquiring a frame image of X-rays incident discretely on a sensitive surface of the image sensor in units of photons; a calculation unit that compresses and stores signal data of pixels having the discriminated significant pixel values, The radiation imaging system according to claim 7 .

9. A radiation imaging system including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and a computer having an image signal processing function for outputting signals electrically converted by the pixels in response to incidence of radiation containing image information as pixel values, The computer included in the radiation imaging system A process of acquiring a frame image of photons or particles of radiation discretely incident on a sensitive surface of the image sensor; A process of adding up pixel values ​​of adjacent blocks of pixels in the image sensor, each block consisting of a plurality of pixels in the vertical and horizontal directions, to obtain a total sum; and The sum is sequentially compared with a threshold value that is predetermined for the sum of pixel values ​​in the block region, and pixel values ​​of each pixel in the block whose sum is lower than the threshold value are removed as pixel data caused by noise inherent to the device, and pixel values ​​of each pixel included in the block whose sum is higher than the threshold value are output as significant pixel signal data. program.

10. A radiation imaging system including an image sensor having a sensitive surface on which a plurality of pixels are two-dimensionally arranged, and a computer having an image signal processing function for outputting signals electrically converted by the pixels in response to incidence of radiation containing image information as pixel values, The computer included in the radiation imaging system A process of acquiring a frame image of photons or particles of radiation discretely incident on a sensitive surface of the image sensor; a process of comparing an output signal for each pixel from the image sensor with a first threshold value that is predetermined as a noise level, and acquiring image signals of pixels having pixel values ​​equal to or greater than the first threshold value; A process of comparing pixel values ​​of pixels having pixel values ​​smaller than the first threshold with a second threshold defined as a noise level lower than the first threshold; a process of generating an image signal by adding a correction value determined by performing arithmetic processing on a plurality of pieces of data acquired in advance to pixel values ​​of pixels having pixel values ​​smaller than the first threshold value and equal to or greater than the second threshold value; and and outputting an image signal of the pixel value of the pixel to which the correction value has been added together with an image signal of a pixel having a pixel value equal to or greater than the first threshold value. program.

11. a process in which the computer dynamically sets the threshold for each individual event of radiation observation to discriminate pixels having pixel values ​​due to noise, and outputs pixel values ​​of significant signals as radiation image data; 10. The program according to claim 9, further comprising: a process of compressing and storing the radiation image data having the significant pixel values.

12. A process in which the computer dynamically sets the first threshold value and the second threshold value for each individual event in which radiation observation is performed, discriminating pixels whose pixel values ​​are caused by noise, and outputting pixel values ​​of significant signals as radiation image data; and compressing and storing the radiation image data having the significant pixel values.

Images