Research on CMOS in Radiation Detection

1 Introduction to CMOS Detector

Radiographic detection technology uses X-rays to detect discontinuities inside materials and displays images on recording media. With the continuous advancement of technology, radiographic detection has expanded from the traditional photographic method using film as the recording medium to a variety of digital radiographic detection methods, such as film digitization, radioscopy, computed radiography, and direct radiography [1]. In practical applications, it is necessary to select the appropriate method based on the resolution and relative sensitivity required for detection. Compared with other radiographic recording media (such as CCD, polysilicon, etc.), CMOS (complementary metal oxide silicon) technology has more performance advantages. At present, the minimum pixel size of CMOS detectors can reach 39μm, with high detection accuracy, good temperature adaptability, and strong structural adaptability.

Compared with the large intensifier imaging system, the CMOS radiographic scanning detector (Figure 1) has a compact structure and high internal chip integration. Compared with the CCD imaging method, each detection point of CMOS has its own amplifier for separate configuration. CMOS converts the received radiation into light through the conversion screen inside, and the detection point unit directly in contact with the conversion screen converts the light into electrons. Each detection point unit has its own amplifier to amplify the electrical signal, and finally performs A/D conversion on the signal in the detector to form a binary code and transmit it to the computer. CMOS is mainly suitable for 20-320 kV radiation energy, 80/μm spatial resolution, 6 lp/mm detection resolution without geometric magnification, and the detection image reaches 4096 gray levels.

Figure 1 CMOS ray scanning detector

2 Detection application of CMOS detector

2.1 Detection process

Since the CMOS ray detection units are arranged in a linear array, only one line of the projection image formed by the rays passing through the inspected object can be obtained in a static state. In order to obtain the image of the inspected object, relative scanning movement is required, and the line-by-line acquisition and splicing are formed into a complete projection image. When obtaining the detection image, the ray energy fluctuation is required to be as small as possible and can work continuously for a long time. Therefore, the author uses a constant voltage ray source (YX-LON MG325, maximum voltage 320 kV, large focus 3.0 mm, small focus 2.O mm). The process of ray detection using CMOS linear X-ray scanning detector is: detector configuration and calibration - determine the penetration mode, adjust the position parameters - relative movement, obtain the scanned image - image processing, defect analysis.

2.2 Design of detection tooling

The imaging unit (linear array) of the detector needs to be well matched with the center line of the ray beam, and relative position tilt and offset cannot occur. Therefore, it is necessary to design a suitable imaging tooling to complete the fixation and position adjustment of the detector and realize the relative movement with the inspection workpiece. The tooling should be able to be easily moved in and out (cylindrical workpieces), and should have certain flexibility and greater adaptability (detection of different types of workpieces).

Based on the principles of simplicity and practicality, the detection tooling is designed on the basis of the existing real-time X-ray imaging system, that is, the detection workpiece is placed on the stage during detection, and can be moved left and right, rotated around the vertical axis, and other movements can be realized; the detector is fixed on the motion axis of the intensifier of the real-time X-ray imaging system through the tooling, and can be lifted vertically and moved forward and backward. In addition, the detector can also be adjusted to a certain angle of rotation. Through the organic combination with the real-time imaging detection system, X-ray detection of various types of workpieces can be realized. In addition, fixed positioning tooling should be designed for the workpiece during application.

2.3 Detector configuration and calibration

When using the detector for the first time, the imager type parameters (length and withstand voltage, etc.) must be specified to determine the minimum available integration time. Before the detector works normally, it must be configured and calibrated so that the bias output and gain output of all detection units are consistent under certain imaging conditions.

For new detection objects, first configure the parameters related to the acquisition image (integration time, scanning accuracy, and whether to superimpose and average), and then start the detector calibration. The effects of focal length and object distance must also be considered during calibration. Generally, three steps are required for calibration: ① Turn off the X-ray source and perform bias calibration on the detector. ② Turn on the X-ray source and adjust the current and voltage values ​​required for detection so that the linear array output signal of the detector reaches the maximum but does not reach saturation. ③ Adjust the X-ray energy so that the linear array output signal is reduced to half of the maximum signal. The calibration results are stored in the form of files and can be used for subsequent detection. However, if the calibration parameters are changed after the call, recalibration is required before detection can be performed.

For most detection objects, the current and voltage values ​​applied during actual detection are relatively high, and the output signal is already saturated when the detector is calibrated. To solve this problem, corresponding calibration test plates are designed according to the detection conditions of different thicknesses. The test plate has uniform thickness. After the first step of calibration is completed, the test plate is placed in the X-ray source window, and then the X-ray is turned on for the next calibration operation.

2.4 Selection of transillumination method

(1) The translation method is suitable for X-ray detection of flat weld workpieces. During detection, the detector and the X-ray source are kept relatively fixed, and the workpiece is placed on the stage and moved parallel to the X-axis at an appropriate speed. For annular welds on tubes and cylinders, if the image is imaged by translation, an elliptical perspective image will be collected. Only the image of the central area can be used for the evaluation of the test results, and it is necessary to rotate at multiple angles to complete the entire test, which reduces the test sensitivity (Figure 2a). In some cases, due to the excessive thickness, the through-illumination test cannot be achieved.

(2) The rotation method requires the relative position to be adjusted so that the workpiece is placed at the rotation center of the stage and is in a straight line with the center of the beam and the center of the detector. For cylindrical parts, the detector is placed inside the workpiece through the tooling, as close to the test site as possible, and the single-wall single-image method is used for through-illumination; for tubular and cylindrical workpieces with smaller inner diameters, the double-wall through-illumination method is used; the through-illumination area can be expanded and imaged by rotating a certain angle, which can effectively improve the test efficiency (Figure 2b). For rotating workpieces, the use of rotational imaging has outstanding advantages, which can improve image quality and shorten the test time. [page]

2.5 Movement speed control

Since the detector must have relative motion to form an image, the movement speed needs to be controlled within a reasonable range. If the speed is not appropriate, the image will be stretched or compressed. In addition, the higher the resolution and the lower the image noise, the lower the movement speed needs to be.

Figure 2 Detection images obtained by different transillumination methods (a) Translation method (b) Rotation method

The moving speed V in translational imaging is related to the exposure time T of the detector, the imaging accuracy P, the transillumination magnification M and the number of repeated scans N.

For the rotation method, the inner diameter of the workpiece needs to be considered for calculation.

2.6 Optimization of detection parameters

The optimal magnification Mopt is related to the inherent unsharpness Us of the detector and the size of the ray focus d [2]:

After calculation, the best magnification Mopt=1, that is, the detector is as close to the workpiece as possible during imaging. In addition, the imaging quality is also related to the selected parameters such as transillumination voltage, current, focal length and focus.

The clarity of the scanned image is related to the number of repeated scans. The Double Graylevel option is used during image scanning, which is similar to the superposition of 4 frames of images (N=4) in real-time imaging detection. The speed of detection is reduced by 4 times, but the image is greatly improved, the noise is significantly reduced, and it is more conducive to the detection and identification of defects. The detected image can meet the AB level requirements specified in the GB 3323-1987 standard.

2.7 Quantitative analysis of defects

When measuring the image size, it is necessary to place the test piece that has been measured or has known precise size close to one side of the weld to be inspected and image it at the same time as the weld. Before each evaluation, a calibration should be performed, and the image size should be compared or converted into the actual size through a formula during defect measurement. For this purpose, a special test piece for measuring and evaluating films is designed (Figure 3), and the test piece can also be used to detect whether the relative motion speed matches.

Figure 3 Test piece for quantitative defect analysis

After the size calibration is completed, the defect quantitative analysis is realized by image processing method. The Canny edge detection algorithm is used to locate the defect edge. Then the detected edge is thinned. Then, by searching for a 5×5 or larger neighborhood centered on the endpoint of each edge line, find other endpoints and fill them, complete the edge point connection, and remove the gaps in the edge detection image. Then, the pixel marking method is applied to check the connectivity of the adjacent points of each target pixel and mark the target within the closed curve. Through the above operations, different defects can be marked for measurement, and finally the defect parameter calculation is completed [3].

2.8 Image archive management

The detection results are stored in the form of digital images on the computer. In order to facilitate the unified management of the detection images, the author designed a management database for image files to record the detection information (workpiece name, detection date, etc.), imaging parameters and detection evaluation results.

3 Application conclusions and problem analysis

CMOS radiation detectors have high spatial resolution (61p/mm, inherent blur <0.2 mm) and high detection sensitivity (4096 gray levels). The imaging quality is better than that of the real-time imaging system using an intensifier, and is close to or reaches the level of film photography; in terms of image contrast, it is better than film photography methods and real-time imaging systems.

Through experimental optimization and other methods, the detector has been successfully applied to the X-ray detection of most product parts such as flat plate welds, circumferential welds and longitudinal welds, which has improved the detection efficiency and reduced the detection cost. In order to better promote the application of digital X-ray detection technology, it is necessary to carry out research work in the following aspects:

(1) Optimization detection and simulation of complex workpieces[4], to provide theoretical support for the interpretation of detection results.

(2) Rapid reading, processing and analysis of large-capacity image files, and research on automated and semi-automated methods for quantitative defect analysis.

(3) Management and transmission of image files (introduction of PACS mode)[5].

(4) Establishment of new digital X-ray detection standards.