RU64797U1 - DEVICE FOR LARGE-SCALE DETECTION OF ROUND-FORM DEFECTS BY X-RAYS OF WELDED COMPOUNDS - Google Patents

Abstract

The device for the multi-scale detection of round-shaped defects by X-rays of welded joints belongs to the field of information technology and can be used in digital image processing systems. In particular, the proposed device can be used in non-destructive testing systems for welded joints of pipelines for the detection of welding defects of round or elliptical shape (pores, slag inclusions) from radiographic images. The utility model is aimed at improving the quality of detecting features of images having a round or elliptical shape (defects in welded joints such as pores and slag inclusions). The specified technical result is achieved by calculating the Shannon entropy of the obtained two-dimensional wavelet images of the analyzed two-dimensional signal and the adaptive choice of scaling coefficients of the wavelet decomposition of the image (x-ray). For this, a block for calculating the Shannon entropy of a two-dimensional wavelet image is added to the proposed device. The resulting entropy values are stored in random access memory and used to calculate the maximum entropy, with variable values of the scaling coefficients and the angle of rotation of the filter of a continuous two-dimensional wavelet transform. When determining the maximum, at the current time, entropy values, discrete values of the wavelet decomposition coefficients, also stored in the random access memory, are output to the device. The maximum values of the obtained sets of wavelet coefficients correspond to the regions of the image features (defective places of the roentgenograms).

Description

The device for the multi-scale detection of round-shaped defects by X-rays of welded joints belongs to the field of information technology and can be used in digital image processing systems. In particular, the proposed device can be used in non-destructive testing systems for welded joints of pipelines to detect welding defects of round or elliptical shape (pores, slag inclusions) from radiographic images.

A promising approach to the analysis and processing of non-stationary signals (which include images), the identification of features and local heterogeneities is the use of continuous wavelet transform and analysis of wavelet spectra [1, 3, 5].

The decomposition of the images on the basis of wavelet functions can be carried out by applying a one-dimensional separable wavelet transform (separately in the rows and columns of the matrix of instantaneous values of a two-dimensional signal) or two-dimensional wavelet transform. To detect the features of rounded or elliptical images, it is more preferable to use a two-dimensional wavelet transform, since it will contain more information about local image inhomogeneities compared to a one-dimensional separable wavelet transform [2-4].

The use of a two-dimensional wavelet transform for image analysis requires the choice of a wavelet, for example, with or without spherical symmetry with a possible input of rotation along with shifts and contractions of the base wavelet [3].

An analysis of the obtained wavelet spectra when detecting image features requires the calculation of a certain quantitative estimate characterizing the wavelet images of the image at different decomposition scales and allowing us to conclude that the detection of features has been completed. Shannon entropy is used as a similar characteristic:

,

where i = 1 ... N, j = 1 ... M are the dimensions of the matrix of wavelet coefficients W, P [W (i, j)] is the probability of the appearance of the corresponding coefficient in the wavelet image, H is the Shannon entropy that allows one to calculate amount of information contained in wavelet coefficients.

The maximum entropy of the wavelet coefficients corresponds to the maximum amount of information contained in the wavelet image and the most reliable representation of the image features in the wavelet space. Thus, determining the maximum entropy of the wavelet images obtained on the given ranges of scaling factors a ₁ , a ₂ , we obtain the optimal representation of the image features in the wavelet space.

A device for computing a continuous two-dimensional wavelet transform with an arbitrary angle of rotation of the filter [Copyright certificate RU No. 60242, IPC 7 G06F 17/14, BI No. 1, 2007], which consists in the formation of a two-dimensional wavelet filter with or without spherical symmetry and rotated at a given angle used for convolution with a two-dimensional signal supplied to the input of the device.

The disadvantage of this device is the inability to quantify the resulting wavelet images and the choice of scaling factors adaptively to the analyzed signal.

This device is selected as a prototype.

The technical result of the utility model is to improve the quality of detecting features of images having a round or elliptical shape (defects in welded joints such as pores and slag inclusions).

The specified technical result is achieved by calculating the Shannon entropy of the obtained two-dimensional wavelet images of the analyzed two-dimensional signal and the adaptive choice of scaling coefficients of the wavelet decomposition of the image (x-ray).

The proposed device that implements an approach to the detection of defects in welded joints based on two-dimensional wavelet filtering of x-rays is shown in the figure, where:

block 1 - x-ray electron-optical Converter (REOP);

block 2 - block calculating a continuous two-dimensional wavelet transform

with an arbitrary angle of rotation of the filter (BVP);

block 3 - Shannon entropy calculation unit (BES);

blocks 4, 5, 7 - random access memory (RAM);

block 6 is a comparator.

The principle of operation of the device is as follows. The x-ray radiation f (λ) transmitted through the weld section is recorded by the REOP (block 1) and converted into an analog signal s (t ₁ , t ₂ ) coming from the output of the REOP to the input of the BWP (block 2), the output of which is discrete the wavelet transform coefficients W (a _i , a _j ) are fed to the BES input (block 3). At the BES output, the Shannon entropy value H (a _i , a _j ) calculated for the current wavelet decomposition coefficients is formed. The entropy value is simultaneously fed to the first input of the comparator (block 6) and the input of RAM1 (block 4). At the second input of the comparator, from the output of RAM2 (block 5), the entropy value Н (а _k , а _l ) is stored in this block on one of the previous clock cycles of the device. From the output of the comparator, the flag f, equal to 1 if H (a _i , a _j ) is greater than H (a _k , a _l ) and 0 otherwise, goes to the control input of RAM1, the output of which, when the value 1 comes to the control input, the input of RAM2 receives the current value of entropy. The flag f also arrives at the control input of RAM3 (block 7), where the current wavelet image of the analyzed image received from the output of the BWP is stored. When the value 1 arrives at the control input of RAM3 at the output of the unit, which is the output of the device, discrete values of the wavelet coefficients W (a _i , a _j ) of the analyzed signal are formed.

Information sources:

1. Vityazev V.V. Wavelet Time Series Analysis: Textbook. - SPb .: Publishing house of St. Petersburg. University, 2001 .-- 58 p.

2. Vorobiev V.I., Gribunin V.G. Theory and practice of wavelet transform. - SPb .: VUS, 1999. - 204 p.

3. Finish I. Ten lectures on wavelets. - Izhevsk: Research Center "Regular and chaotic dynamics", 2001. - 464 p.

4. Malla S. Wavelets in signal processing: Per. from English - M .: Mir, 2005 .-- 671 p., Ill.

5. Pereberin A.V. On the systematization of wavelet transforms // Computational methods and programming. - 2001 - T. 2. - S.15-40

Claims (1)

A device for calculating a continuous two-dimensional wavelet transform with an arbitrary angle of rotation of the filter, containing blocks that generate a two-dimensional wavelet filter and two-dimensional convolution of the analyzed signal with this filter, characterized in that it contains an x-ray electron-optical converter, to the input of which the x-ray radiation f (λ), passed through the weld section, and an analog signal s (t ₁ , t ₂ ) is generated at the output, the input unit for calculating the continuous two-dimensional ivelet transform, from the output of which the discrete values of the wavelet transform coefficients W (a _i , a _j ) go to the input of the third random access memory and the input of the Shannon entropy calculation unit, the output of which forms the Shannon entropy value Н (а _i , а _j ) calculated for the current coefficients of the wavelet decomposition, which simultaneously enters the input of the first random access memory and the first input of the comparator, the second input of which is output from the second random access memory AET H (a _k, a _l) entropy value, stored in this block at a previous cycles of the device, and the output comparator flag f, equal to 1 when H (a _i, a _j) more than H (a _k, a _l ) and 0 otherwise, it goes to the control input of the first random access memory, from the output of which, when the value 1 is received at the control input, the current entropy value goes to the input of the second random access memory, the flag f also goes to the control input of the third memory, which output is At the device input, when the value 1 is received at the control input, discrete values of the wavelet coefficients W (a _i , a _j ) of the analyzed signal are formed at the maximum, at the current time, Shannon entropy value.