KR20110017204A - Method for restorating geometric radiography of boiler tube welds - Google Patents

Abstract

PURPOSE: A method for restoring the geometric radiography of boiler tube welds is provided to accurately detect the defects such as minute distortion or noise within a geometric radiography of boiler tube welds. CONSTITUTION: A method for restoring the geometric radiography of boiler tube welds comprises the steps of: collecting a geometric radiography of boiler tube welds; restoring an image for an inner region of a divided boiler tube through geometric structure analysis to which a function fitting method is applied; selecting pixel lines of a right angle one by one, wherein the pixel lines exist in the geometric radiography; allowing the selected pixel line to be matched with one of polynomial, Gaussian, prototype and elliptic functions; and accurately restoring an image by quantitatively judging the image distortion of geometric radiography.

Description

METHOD FOR RESTORATING GEOMETRIC RADIOGRAPHY OF BOILER TUBE WELDS}

The present invention relates to a method for geometrically reconstructing a radiographic image, and more particularly, to a method for reconstructing geometrically precisely finding and minimizing defects such as fine distortion or noise in a radiographic image.

In general, thermal power plants have about 40,000 boiler tube welds, of which only one weld is damaged, which is a serious situation that requires the entire plant to be shut down. For this reason, all boiler tubes are undergoing digital radiography to check for abnormalities in the welds of boiler tubes. Digital radiography includes direct digital radiography (DDR) by semiconductor sensors and computed radiography (CR) by which images are acquired by optically stimulating image plates. Penetration inspection by direct digital radiation and indirect digital radiation can reduce inspection time by 1/10 compared with conventional film method, and it does not develop processing like film, so it can be applied to other types of film method. It is applied in all industries in a very effective way.

An image acquired by such digital radiography is referred to as a radiographic image, wherein the radiographic image acquires an image by passing the radiation through a boiler tube, but a series of processes absorbed by the acquisition medium is dependent on the geometrical conditions of the photographing conditions. Accordingly, there is a problem in that the prototype of the boiler tube cannot be reproduced as it is distorted by the nonlinear element.

In addition, the high energy of the radiation used in the radiographic test results in transmission, scattering and reflection, which generates noise in the transmitted image. That is, the radiographic image shown in front of the examiner evaluating the radiographic image includes a distorted and noisy image unlike the actual image. Therefore, it is impossible to accurately detect and repair fine defects such as distortion and noise present in the welded portion of the boiler tube by current radiographic test.

The present invention is to solve the above-mentioned problems, by removing the noise of the image and correcting the image distortion to present accurate image data and at the same time minimize the effect of noise through the geometric composition analysis as reliable image It is an object to provide a method for restoring.

In order to achieve the object of the present invention as described above, and to perform the characteristic functions of the present invention described below, features of the present invention are as follows.

According to an aspect of the present invention, the radiographic image of the boiler tube is collected, and the image is restored by a geometric composition analysis to which a function-matching method is applied to the internal region of the divided boiler tube, wherein the geometric composition analysis is the radiation transmission By extracting the pixel lines in the direct direction of the image one by one and matching any one of the functions of multi-order function, Gaussian function, circular function and elliptic function, the matching function with the lowest matching error is extracted and based on the result, A method for accurately reconstructing an image by judging the distortion systematically is provided.

Here, the geometric composition analysis, (a) calculating the longitudinal inclination in the line pixel arrangement in the longitudinal direction of the radiographic image inside the tube involved in the weld, (b) the image area of the non-welded-welded-non-welded Comprising a composition analysis in the radial direction of the line pixel array based on the segmentation to obtain a matching function with a small matching error, (c) at the same time flattening the distortion of the image based on the parameters obtained in the steps (a) and (b) Realizing an image, and (d) maximizing visual effects by region selection and differential image processing on the planarized image data.

In addition, in the step (a), the noise is removed from the line pixel arrays arranged in the longitudinal direction of the radiographic image using an average filter, and the corresponding function is applied to each line pixel array from which the noise is removed. Matching is performed to calculate the slope mean and standard deviation in each array.

Also, in the step (b), the segment consisting of the non-welding-welding-non-welding is divided based on the standard deviation of the inclination, so that the pixel group in the welded portion in the radial line array is excluded from the compositional analysis. After the noise is removed by applying an average filter to the pixel array in the radial direction of the non-contact portion, function matching is performed on each pixel array, the mean and the standard deviation are calculated, and the matching function can be obtained from the average.

In addition, in step (c), a pattern obtained by geometrically shaping the non-welded portion excluding the welded portion from the radiographic image obtained at the time of measurement based on the results of the steps (a) and (b) is made, and the pattern is circled. It can be applied to an image to correct distortion of the image and at the same time implement a planar region image.

Also, in the step (d), the image of the non-welded portion may be excluded from the image enhancement by masking the data of the planar region image, and the image enhancement may be performed on the joint portion of the welded portion and the welded portion and the non-welded portion. Can be.

According to another aspect of the present invention, there is further provided a computer readable recording medium having recorded thereon a computer program for executing the above method.

According to the present invention, by correcting the radiographic image of the boiler tube through the geometric composition analysis to which the water-containing method is applied, it is possible to restore the image with high accuracy and to contribute to the safe operation of the thermal power plant.

In addition, according to the present invention, the restoration method through the geometric composition analysis with the functional bonding method can be applied to all digital radiography images used in the field as well as the boiler tube of the power plant, so that the economic profit is greatly generated. Effect is achieved.

The following detailed description of the invention refers to the accompanying drawings, which illustrate, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It is to be understood that the various embodiments of the invention are different, but need not be mutually exclusive. For example, certain shapes, structures, and characteristics described herein may be embodied in other embodiments without departing from the spirit and scope of the invention with respect to one embodiment. In addition, it is to be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the invention. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present invention, if properly described, is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled. Like reference numerals in the drawings refer to the same or similar functions throughout the several aspects.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention.

Example of restoration of radiographic images

In the method for geometrically restoring a radiographic image according to an embodiment of the present invention, structural analysis of shape generation according to radiation absorption and transmission of a boiler tube should be preceded. That is, the radiation emitted from the radiation source in the measurement composition of the radiation source-boiler tube-detection plate is proportional to the square of the distance by the inverse square law when the source is in the form of point in the progress in free space. The strength is lowered and the strength is exponentially lowered in the medium by Beer's law. The shape, thickness, and absorption constant of the boiler tube 100 are added to the measurement conditions or the environment to determine the absorption and transmission patterns, and the results are shown in the image. In this process, the circular structure of the inner diameter and the outer diameter is reflected on the radiographic image together with the base portion, the thickness portion, and the inner region of the tube 100. The relationship between the thickness of the base portion and the inner region of the tube 100 may be illustrated in FIG. 1.

1 is a view showing the area division by the radiographic test of the boiler tube 100 to be used in the present invention, the thickness of the base portion and the inner region of the tube 100 is visible to the naked eye and different from the thickness portion Thickness measurement is possible by the difference in chromaticity between. In the outermost part of the three areas, the background part is completely exposed to radiation, and is excluded from the reading. However, in the case of a film image, unevenness of film particles causes color noise and noise is generated. The second area, the thickness part, is used to measure the thickness in the radiographic image where the thickness increases due to the incident radiation and the area where the increase occurs most. The inner region of the boiler tube 100 is a region requiring radiographic inspection to remove distortion and noise in the present invention, which is divided into a welded portion and a non-welded portion, and serves as an image reading to determine a welding state after a welding operation. In addition, even if not a welding section also serves to provide information for identifying the state of the cracks, erosion, corrosion and rupture inside the tube 100.

In the inner region of the tube 100, the relationship between the dose of radiation before and after the transmission of the medium by the law of beer and the inverse square law physically predicts the degree of photosensitization of the image detection plate, and the radiation source including the geometric distance calculation and the intersection and angle of incidence. Although an image forming model may be considered that combines the separation distance, thickness, absorption constant, density, and imaging sensitivity of the image plate from the tube 100, the principle of the point source, which is a premise of the inverse square law, is due to the irregularity of the field measurement environment. As it is not observed, there is a limit in applying the Beer's law according to the change in the thickness of the circular bend. In addition, it is difficult to accurately measure the incident radiation dose because high energy radiation intensity above 150 KVP must be adjusted from time to time. That is, it is not practical to obtain a mathematical analysis model by applying physical laws as they are. In order to overcome these limitations and to investigate compositional distortion and noise generation due to radiation transmission, it is desirable to obtain an analytical model through measured image analysis rather than a mathematical analysis physical model.

In this regard, in an embodiment of the present invention, first, a radiographic image of the boiler tube 100 is collected, and a geometrical structure in which a function fitting method is applied to the inner region of the divided boiler tube 100 is applied. The analysis will restore the image.

Here, the geometric analysis analyzes the pixel lines in the direct direction of the radiographic image one by one and matches any one of a multi-order function, a Gaussian function, a circular function, and an elliptic function to extract a matching function with the lowest matching error. Distortion of the transmitted image can be determined regularly, and the image can be accurately restored. The multi-order function, Gaussian function, circular function, and elliptic function can be expressed as the following equations (1), (2), (3), and (4). In this case, a least square error summation is used as an error discrimination method.

A ₁ , b ₁ , c ₁ , a ₂ , b ₂ , c ₂ , a ₃ , b ₃ , c ₃ , a ₄ , b ₄ , c ₄ are constants, and t is a variable.

On the other hand, since the error occurring in the radiographic image is mainly caused by the noise of the image, when the matching error described above corresponds to the smallest matching function, the matching function exhibits a powerful analysis power not only in shape composition but also in noise characteristics. . Hereinafter, the geometric composition analysis method will be described in more detail.

Geometric composition example

2 is a flowchart illustrating a geometrical analysis method according to an embodiment of the present invention.

As shown in FIG. 2, in the geometric composition analysis method S200 according to an embodiment of the present invention, (a) calculating a longitudinal gradient (S210) and (b) analyzing the composition in a radial direction (S220). ), (c) image correction and planarization (S230), and (d) maximizing visual effects by region selection and differential image processing (S240).

First, in the step (a) of the present invention, the radiographic image of the inside of the tube 100 in which the welding part is interposed may be represented by the following equation (5) in the longitudinal line pixel arrangement in the radiation-proof image. Noise is removed by applying an average filter of d (k, i), which is the number of elements N of the moving window.

In Equation (5), k and i represent spatial coordinate positions on the pixel array, and j represents the number of moving windows.

Subsequently, in step (a) of the present invention, the linear function analysis is performed on each line array of which noise is removed to calculate a slope average and a standard deviation in each array, thereby performing geometric composition analysis.

Subsequently, in the step (b) of the present invention, the composition of the pixel group in the welded portion of the radial line pixel array is divided by dividing the section consisting of the non-welded portion, the welded portion, and the non-welded portion based on the standard deviation of the slope obtained in the step (a). Except in, the noise is removed by applying an average filter to the pixel array in the radial direction of the non-welded part, and the function is matched to each pixel array, and then the average and the standard deviation are calculated. The geometrical analysis is performed by finding. In this case, the representative matching function refers to a matching function having the smallest matching error as described above. The geometrical composition of the section region and the inclination of the non-weld-weld-non-weld described in the above steps (b) and (a) may be shown as shown in FIG. 3.

The graph shown in Figure 3 shows the region formation of the pixels on the vertical axis and the horizontal axis of the boiler tube radiographic image.

Reference numeral 310 denotes a pixel region divided into a vertical axis and a horizontal axis in the boiler tube radiographic image. The upper graph of 320 denotes an area of the vertical axis in gray level and a lower graph of an area of the horizontal axis in gray level. As shown by reference numeral 320, the difference according to θ of the gray level is shown, and thus, the radiographic image is geometrically distorted.

Returning to step (c) of the present invention, the non-weld part except for the weld part is geometrically obtained from the radiation source image obtained at the time of measurement based on the geometric parameters of the length and the radial direction calculated in the steps (a) and (b). A regularized pattern is created, and the pattern is applied to the original image to correct distortion of the image, and at the same time, a planarized image (plane area image) is realized.

Subsequently, in the step (d) of the present invention, the image of the non-welded part is excluded from the image enhancement by masking the data of the planar region image implemented in the step (c), and the joint part of the welded part and the welded part and the non-weld part is removed. Perform image enhancement on.

Example of function matching

4 is an analysis graph exemplarily showing the degree of function fitting by randomly extracting 17 indirect digital radiographic image samples according to an exemplary embodiment of the present invention.

E _j and γ _j shown in FIG. 4 correspond to a matching error with xl (t), which is image data of a linear linear pixel array according to matching of the matching functions f _j (t) and (j = 1,2,3,4). In this regard, the formulas (6) and (7) can be represented as follows.

As in equations (6) and (7), X _l , F _j is a determinant representation of x _l (t) and f _j (t), and μ _l = mean (Xl). In addition, e _j is an error sum mean and γ _j is a least square error summation. The e _j and γ _j are measures of approximation and interpolation of f _j (t) with respect to x _l (t), respectively, and as e _j approaches 0, x _l (t) The graph form of is approximately coincident with the function trajectory of f _j (t), and as γ _j approaches 1, f _j (t) is able to interpolate pixel values of x _l (t) with near error. will be. When noise is mixed in the radiographic image, e _j and γ _j differ in propensity according to the variance of the noise distribution. That is, even though e _j is close to 0, γ _{j 〈} 1 may be sufficient if the noise variance value is relatively large, and e _j ≒ 0 and γ _j ≒ 1 may be satisfied simultaneously when the variance value is extremely small. In addition, when γ _j of the corresponding image approaches 1 for images satisfying e _j ≒ 0, the noise variance value may be small, and when γ _j ≪ 1, the noise variance value may be relatively high.

Meanwhile, in FIG. 4, when _j performs a function match of a fifth order function among multiple orders, e _j and γ _j of the function match and the relation are shown as 17 CR image samples. The reason for using a fifth order function but fully described in the subsequent Fig. 5, the image transmission in accordance with the information Fig meet's function can sometimes take a variety of functions, such as circular, oval or infinite circular structured satisfying e _j ≒ 0 This is because there may be no representative function.

As such, in the present embodiment, as a result of considering γ _j characteristics based on the commonality of e _j ≒ 0, e _j is constant as 0 even if there is a difference in the value of γ _j which shows noise variance according to function matching. can confirm. That is, although irregular noise of the radiographic image due to various radiographic experiment environments is generated, the image data of the boiler tube 100 according to the present embodiment proves that geometric matching patterns can be extracted. Function Match Result

5A to 5C are diagrams exemplarily illustrating an analysis of an actual function matching result according to an embodiment of the present invention. FIG. 5A is a graph related to elliptic function matching, and FIG. 5B is a graph related to original function matching. 5C is a graph relating to infinity radial circle function matching. As shown, the graph of each analysis result shows that the image is generally darkened or lightened depending on its intensity when irradiated with radiation to measure the weld of the boiler tube 100, but irrespective of such radiation energy. The image shows that the original geometry of the boiler tube 100 is preserved.

Conventional radiographic readings have been mostly relying on visual inspection of the imaged results, and in most cases, re-testing was performed in the case of over-permeation or under-permeability, which is difficult to visually identify. The analysis of the method reveals that the image data itself is preserved in geometric composition, regardless of whether it is excessive or insufficient radiation transmission.

6 is performing the function, consistent with the image data 444 of boiler tubes (100) in accordance with one embodiment of the present invention and a graph showing a result of actual measurement of the error by way of example, a unit ^10-13 as e _j axis Shows near zero. Through these facts, it is fully demonstrated that the geometric shape is preserved in the radiographic image data regardless of distortion or noise generation due to the welding radiation environment of the boiler tube 100.

7A to 7G are diagrams exemplarily illustrating a noise removing process according to an embodiment of the present invention as image data, and FIG. 7A illustrates straight line image data including noise, where A is a normal part and W is a welded part. Indicates. FIG. 7B illustrates a data form from which noise is removed by the average filter in step (a) of FIG. 2, FIG. 7C illustrates an image data array from which noise is removed as a result of distortion correction by gradient estimation, and FIG. 7D illustrates noise data. Represents an array. 7E to 7G show the measured radiographic image, the noise-removed image, and the noise image, respectively. Through the above process, in the present embodiment, it can be seen that image data and distortion patterns corresponding to the circles and noise of the radiographic image of the boiler tube 100 can be classified.

FIG. 8 is a diagram illustrating a result of correcting a measured image having distortion, noise, and an error according to an embodiment of the present invention, and extracting geometric factors of a radiographic image structure in which noise and distortion exist. Through the first correction, the measured image data is flattened, and after that, image processing of the normal part and the weld part is performed differently, and thus only the weld part is selectively image-improved, thereby maximizing the visual effect.

Accordingly, in the present embodiment, the original pattern of the boiler tube 100 can be restored by applying the mathematical pattern extraction, correction, and image enhancement method inherent in the image data while minimizing the influence of the radiation measurement environment of the welding portion of the boiler tube 100. You will know that you can.

As described above, the method according to an exemplary embodiment of the present invention may be implemented in the form of program instructions that may be executed through various computer components, and may be recorded in a computer-readable recording medium. The computer-readable recording medium may be understood to include program instructions, data files, data structures, etc. alone or in combination.

Program instructions recorded on the computer-readable recording medium may be those specially designed and configured for the present invention, or may be known and available to those skilled in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD-ROMs, DVDs, and magneto-optical media such as floptical disks. media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the process according to the invention, and vice versa.

Figure 1 is a view showing the area division by the radiographic test of the boiler tube 100 to be used in the present invention.

2 is a flowchart illustrating a geometrical analysis method according to an embodiment of the present invention.

FIG. 3 is a diagram exemplarily illustrating a geometric structure of a section region and a slope according to an embodiment of the present invention. FIG.

4 is an analysis graph exemplarily showing the degree of function fitting by randomly extracting 17 indirect digital radiographic image samples according to an exemplary embodiment of the present invention.

5A to 5C are diagrams exemplarily illustrating an analysis of an actual measurement result of an optimal function match according to an embodiment of the present invention.

FIG. 6 is a graph illustrating a result of performing a function matching on image data of 444 boiler tubes and measuring an error thereof according to an embodiment of the present invention.

7A to 7G are diagrams illustrating image removal processes, according to an embodiment of the present invention.

8 is a diagram illustrating a result of correcting a measurement image having distortion, noise, and an error according to an embodiment of the present invention.

Claims (9)

Collect the radiographic image of the boiler tube, and restore the image by the geometric composition analysis applied to the inner region of the partitioned boiler tube, The geometric composition analysis, The pixel lines in the direct direction of the radiographic image are extracted one by one to match any one of the functions of the multi-order function, the Gaussian function, the circular function, and the elliptic function to extract the matching function having the lowest matching error, and based on the result Method for accurately reconstructing the image by determining the distortion of the transmission image. The method of claim 1, The geometric composition analysis, (a) calculating longitudinal gradients in the arrangement of the line pixels in the longitudinal direction of the tube inner radiographic image in which the weld is involved; (b) analyzing the composition in the radial direction of the line pixel array based on segmentation of the image area of the non-weld-weld-non-welded portion to obtain a matching function having a low matching error; (c) realizing a flattened image simultaneously with correcting distortion of the image based on the parameters obtained in steps (a) and (b); and (d) maximizing a visual effect by region selection and differential image processing on the planarized image data; Method comprising a. The method of claim 2, The step (a), The noise is removed from the line pixel arrays arranged in the longitudinal direction of the radiographic image using an average filter, and the corresponding function is matched to each line pixel array from which the noise is removed to obtain a gradient in each array. Calculating the mean and standard deviation. The method of claim 3, In step (b), The pixel array in the radial direction of the non-welded portion is excluded from the compositional analysis by dividing the section consisting of the non-weld portion-weld portion-non-weld portion based on the standard deviation of the gradient, except for the composition analysis of the pixel group in the radial line array. Applying a mean filter to remove the noise, performing a function match on each pixel array, calculating the mean and the standard deviation, and calculating the match function using the mean. The method of claim 4, wherein In step (c), On the basis of the results of the steps (a) and (b), a pattern obtained by geometrically shaping the non-welded parts excluding the welded part from the radiation original image obtained at the time of measurement, and applying the pattern to the original image to remove distortion of the image Correcting and simultaneously implementing a planar region image. The method of claim 5, The step (d) And removing the image of the non-welded portion from the image enhancement by masking the data of the planar region image, and performing image enhancement on the joint portions of the welded portion, the welded portion, and the non-welded portion. The method of claim 2, The matching error is represented by the sum of the matching errors as shown in Equation (6) below, and when the matching error is determined using the least error sum of Equation (7), As e _j approaches 0, the graph form of x _l (t) approximately matches the function trajectory of f _j (t), and as γ _j approaches 1, f _j (t) becomes x _l ( and interpolating the pixel value of t) with a near error.

Wherein f _j (t), (j = 1,2,3,4) is a matching function, and x _l (t) represents image data of a linear array of radial pixels. The method of claim 2, The average filter, Equation (5) is applied, characterized in that the noise in the pixel array is removed by the application of equation (5).

K and i represent spatial coordinate positions on the pixel array, and j represents the number of moving windows. A computer-readable recording medium having recorded thereon a computer program for executing the method according to any one of claims 1 to 8.

Images