RU2828190C1 - Method for non-destructive testing of barrel of small arms and cannon weapons by computer vision - Google Patents

Abstract

FIELD: non-destructive testing.

SUBSTANCE: invention relates to research and diagnostics of firearms, namely to a method for non-destructive testing of a barrel of small arms and cannon weapons. Method can be used in solving military tasks. Non-destructive inspection of the barrel bore is carried out along its entire length or on separate sections by obtaining images of its surface with a camera with further processing by a computer vision system, which consists of a convolutional neural network for detecting and classifying defects and a unit for spatial marking thereof, for calculating geometric and optical metrics, with output of the following results: map of defects of the barrel of small arms and cannon weapons on the digital images of the scan of its full cut, the number of areas for seven different defects, the total area of each defect, number of areas for each defect, having high, moderate and low compactness and minimum, medium and maximum depth.

EFFECT: possibility of carrying out in any conditions of area (laboratory, workshop, polygon, et cetera) of non-destructive testing of channels of rifled and smooth-bore small arms and cannon weapons and determining the bore channel defects map on its full slice scanning digital images, the number of areas for seven different defects, the total area of each defect, number of areas for each defect, having high, moderate and low compactness and minimum, medium and maximum depth, by means of a simple software and hardware system with universal equipment by an unskilled (without special training) operator.

3 cl, 6 dwg

Description

The invention relates to the field of research and diagnostics of firearms, namely to a method of non-destructive testing of the bore of small arms and guns (SAG). The method can be used to solve military problems.

An analysis of the existing level of technology in the specified area has shown that quite a lot of methods of non-destructive testing of the barrel bore are known. The closest in technical solutions to the proposed invention are the methods implemented by various systems and devices for non-destructive testing of the barrel bore, which are described in patents RU 2251072 C2, RU 2494368 C1, RU 2494369 C1, RU 2498266 C1, RU 2368885 C2, RU 113829 U1 and others. They differ, respectively, in the methods of monitoring the wear of the barrel bore, the considered physical and technical parameters of the firearms (FA) and (or) small arms and guns (SAG), on the basis of which a decision is made on the degree of wear of the barrel bore.

The closest in terms of the technical result achieved and having a purpose coinciding with the purpose of the proposed invention (prototype) is the method described in patent RU 2251072 C2. For its implementation, a "System for inspecting defects of the inner surface and measuring wear of the bores of rifled barrels" is used, consisting of a probe with a head for inspecting the bore and a video camera, a power supply and control unit with a controller, an electronic computer (EC), a monitor, an adapter cassette, a movable probe made in the form of a long thin-walled cylinder on which a measuring head with a ring level is located, suspensions of the front and rear wheels connected to movable and fixed bushings located on the thin-walled cylinder. Before measuring the wear of the bore, the measuring probe is calibrated in a special gauge. During the operation of the system, the computer calculates the coordinates of the peg holes, compares them with the passport values obtained during a special calibration of the measuring head. The computer sends a command via the controller to move the probe along the barrel, simultaneously measuring the barrel diameter along the lands and grooves with pins and inspecting its inner surface. The inspection is performed by rotating the video camera sequentially through six positions at 60°. The computer compares the results obtained with the results obtained during calibration, calculates barrel wear and displays a detailed picture of the condition of the inner barrel surface (by cavities, cracks, cavities, corrosion, etc.) on the monitor display. The prototype is illustrated by several examples with different design features and structures of measuring heads, suspensions and links.

The disadvantages of the prototype are:

impossibility of use for inspection of defects of the internal surface and measurement of wear of smoothbore barrels;

the use of a large number of complex, high-precision, specialized and narrowly focused elements (measuring probe, special caliber, etc.), on which the measurement result directly depends;

labor-intensive work with complex equipment, requiring special operator training.

The objective of the invention is to create a method for non-destructive testing of a barrel bore based on a universal and easy-to-use hardware and software complex for analyzing the degree of damage and wear of the bores of rifled and smooth-bore SPV barrels of any caliber.

The essence of the invention consists in carrying out non-destructive testing of the bore along its entire length or in individual sections by obtaining images of its surface with a camera with their subsequent processing by a computer vision system, which consists of a convolutional neural network (CNN) for detecting and classifying defects and a block for their spatial marking, calculating geometric and optical metrics (GOM), with the output of the following results: a map of defects of the bore of the SPV on digital images of the scan of its full section, the number of areas for seven different defects, the total area of each defect, the number of areas for each defect having high, moderate and low compactness, and the smallest, average and greatest depth.

The technical result of the claimed invention consists in the possibility of carrying out non-destructive testing of the barrel bores of rifled and smooth-bore SPV by an unskilled (without special training) operator in any terrain conditions (laboratory, workshop, testing ground, etc.) and determining a map of the barrel bore defects on digital images of the scan of its full cut, the number of areas for seven different defects, the total area of each defect, the number of areas for each defect having high, moderate and low compactness, and the smallest, average and greatest depth, by means of a simple software and hardware complex with universal, widespread and accessible equipment.

The technical result of the claimed invention is achieved by consistently performing measures to prepare and conduct non-destructive testing of the bore of the SPV by using a hardware and software complex for non-destructive testing of the bore using the computer vision method (HSC) (the HSC is schematically shown in Fig. 1, where: 1 - a probe with a digital camera; 2 - a housing for the device for input, output and rotation of a probe with a digital camera (UVVP); 3 - a ring for centering the probe with a digital camera in the bore; 4 - stepper motors; 5 - guide wheels of the stepper motors; 6 - a controller board; 7 - a nylon tape; 8 - a barrel of a small arms and cannon armament; 9 - U VVP; d - SPV caliber, l _st - barrel length; L - the distance from the outer edge of the UVVP to the muzzle end of the bore), which consists of a probe with a digital camera (1 in Fig. 1), UVVP (2 and 9 in Fig. 1), replaceable centering rings a probe in the barrel bore (3 in Fig. 1) and a personal computer (PC) with installed specialized software (SPO), with the help of which the probe movement is automatically controlled, images of the barrel bore surface are obtained and analyzed, with the subsequent output of the following results: a map of barrel bore defects SPV on digital images of the scan of its full cross-section, the number of areas for seven different defects, the total area of each defect, the number of areas for each defect having high, moderate and low compactness, and the smallest, medium and greatest depth (the PAC architecture is schematically shown in Fig. 2). In this case, the following requirements are imposed on the device for obtaining and transmitting the image (probe with digital camera): the diameter of the probe with digital camera is less than the caliber of the firearm being examined and diagnosed; the length of the probe with digital camera is not less than the length of the barrel bore of the firearm being examined and diagnosed; the presence of a built-in light source to eliminate distortions caused by extraneous illumination; the ability to take color photographs at an angle of 90 degrees relative to the probe's movement with a resolution of the resulting images of at least HD (high definition) and a size of at least 1280 by 720 pixels.

In preparation for non-destructive testing of the SPV barrel bore: if necessary, the firearm is cleaned (in accordance with the procedure established by the regulatory and technical documentation); any installed devices (for example, the muzzle brake-compensator) are removed from the muzzle of the barrel bore; the PAK is deployed, in particular, onto the probe with a digital camera (1 in Fig. 1), if necessary, a ring for its centering is installed in the barrel bore (3 in Fig. 1) with a diameter 0.2 mm smaller than the caliber of the SPV, the UVVP (2 and 9 in Fig. 1) is tightly secured with its outer edge using a nylon tape (7 in Fig. 1) on the muzzle end of the barrel bore (8 in Fig. 1), the PC is turned on and the SPO is loaded, into which data are entered on the caliber (d in Fig. 1) and the length of the SPV barrel bore (l _st in Fig. 1) in millimeters, the distance from the outer edge of the UVVP to the muzzle end of the barrel bore (L in Fig. 1) in millimeters (by default, this parameter is set to 0 mm in the SPO, i.e. L=0 mm), the camera shooting sector of the barrel bore surface for the specified caliber of the SPV in degrees, and data on sections of the barrel bore on which non-destructive testing is carried out (in particular, the distance from the edge of the breech of the barrel to the beginning of the sections in millimetres and their length in millimetres).

As an explanation, the following should be noted:

1. The outer edge of the UVVP (2 and 9 in Fig. 1) is the initial (zero) mark from which the distance (L in Fig. 1) of movement of the probe with the digital camera (1 in Fig. 1) is counted (by default, 0 mm is set in the SPO on the PC, i.e. L=0 mm), therefore it is secured tightly to the muzzle end of the SPV barrel bore (8 in Fig. 1). If it is impossible to secure (place) the UVVP using a nylon tape (7 in Fig. 1) in the specified way (for example, due to the presence of a muzzle brake-compensator on the muzzle end of the barrel bore), the distance from the outer edge of the UVVP to the muzzle end of the SPV barrel bore (L in Fig. 1) is manually entered into the SPO on the PC.

2. By default (set in the SPO on the PC), non-destructive testing is performed for a full cut (360 degrees) of the inner surface along the entire length of the SPV barrel bore. If it is necessary to perform non-destructive testing of certain sections of the SPV barrel bore, the following data are manually entered into the SPO on the PC for each of these sections: the distance from the cut of the breech of the barrel to the beginning of the section in millimeters and its length in millimeters.

3. The PAK includes a set of replaceable rings for centering the probe with a digital camera in the barrel bore (3 in Fig. 1) with different diameters from 5.25 to 151.8 mm (0.2 mm less than the main SPV calibers).

4. For ease of operation, depending on the conditions of non-destructive testing and the size of the firearm, the SPV is placed on a horizontal surface (as a rule, for SPV samples with a caliber of no more than 14.5 mm), the PAK is deployed next to the muzzle.

The non-destructive testing of the SPV barrel bore (8 in Fig. 1) begins by pressing a button in the SPO on the PC, after which the UVVP (2 and 9 in Fig. 1), having received a signal from the PC to the built-in control controller (6 in Fig. 1), starts the movement of the probe with a digital camera (1 in Fig. 1) and its centering ring in the barrel bore (3 in Fig. 1) (ZKK) to the edge of the breech of the barrel bore, i.e. to a distance equal to the sum of the length of the SPV barrel bore and the distance from the outer edge of the UVVP to the muzzle edge of the barrel bore (l _st + L). Then the ZKK moves back, shooting a full cut (360 degrees) of the inner surface of the barrel bore. In this case, if the data on the sections requiring non-destructive testing (the distance from the breech face of the barrel to the beginning of the section and its length) are not manually entered into the PC-based SPO, then the complete cross-section of the inner surface of the SPV barrel bore is taken along its entire length from the breech face to the muzzle face. If the data on the sections requiring non-destructive testing are entered into the PC-based SPO, then the complete cross-section of the inner surface of the SPV barrel bore is taken only in these sections sequentially as the SPV moves from the breech face to the muzzle face. The sequentially obtained images are automatically transferred to the PC-based SPO, where they are used to form a scan of the complete cross-section of each section.

As an explanation, the following should be noted:

1. The input, output and rotation of the ZKK are carried out by using guide wheels on two stepper motors (4 and 5 in Fig. 1) in the UVVP, one of which is responsible for the movement of the ZKK along the axis of the barrel bore, the second - for the rotation of the ZKK around its axis.

2. Input, output and rotation of the ZKK, as well as obtaining images of the surface of the barrel bore cut is carried out in the automatic mode of the SPO by using data on the caliber (d in Fig. 1) and the length of the SPV barrel bore (l _st in Fig. 1), the distance from the outer cut of the UVVP to the muzzle cut of the barrel bore (L in Fig. 1) a, the sector of shooting by the camera of the surface of the barrel bore for the specified caliber of the SPV, as well as data on the sections of the barrel bore on which non-destructive testing is carried out (in particular, the distance from the cut of the breech of the barrel to the beginning of the sections in millimeters and their length in millimeters).

Detection, classification and definition of defect boundaries in digital images of the full bore scan (along its entire length or in individual sections) is performed by the SPO using the YOLO architecture CNN, which allows identifying common features in various bore images after training based on a previously collected database and using the derived patterns to work with new data. To adapt the YOLO architecture to non-destructive testing of the bore, the CNN was trained based on the data of the 7 most common bore defects shown in Fig. 3 (determined within the framework of the conducted research work):

- carbon deposits - oxidized deposits remaining on the surface of the barrel bore after exposure to high temperatures on the maintenance equipment of the SPV (oils, lubricants, special liquids, etc.);

- cavities - significant depressions in the metal on the surface of the barrel bore, formed as a result of progressive wear of the barrel and careless maintenance of the SPV (formed as a result of long-term corrosion in places where chrome has chipped);

- corrosion - a surface defect in the form of spots or stripes with a loose structure of the oxide film, formed as a result of water (moisture) entering the barrel;

- a burn grid - a defect in the form of intersecting stripes or thin lines on the surface of chrome-plated barrels that appears during shooting (as the number of shots increases, cracks form in these stripes, and the chrome begins to chip in the form of dots, which can lead to chrome chipping);

- chrome chip - the most pronounced break in the chrome coating of the barrel bore;

- rash - primary lesion of the surface of the barrel channel in the form of dots and small specks;

- scratches - a defect on the surface of the barrel bore in the form of lines (sometimes with a noticeable rise in the metal along the edges), formed as a result of dirt particles (dust, sand grains, etc.) getting into the barrel.

The result of the SNS operation is a map of the barrel bore defects on digital images of the scan of its full section (along the entire length of the barrel or on its individual sections), which is transferred to the BRM, where, based on the boundaries of the barrel bore defects, their geometric and optical metrics are calculated in accordance with the formulas presented in the table in Fig. 4. As a result of the BRM operation, for each of the 7 barrel bore defects the following are determined:

- the number of defect areas in pieces (i, pieces);

- the total area of the defect as a percentage (S _total , %) relative to the total area of the full bore scan obtained during non-destructive testing (along the entire length of the barrel or in its individual sections), which is calculated using the formula

where S _p is the area of the full bore scan obtained during non-destructive testing (along the entire length of the bore or in its individual sections); S _i is the area of the defect in the i-th region;

- the number of defect areas with high, moderate and low compactness (characterizes the density and locality of the defect), in pieces (i _q , pieces), which are determined by the following rules: if the compactness of the defect in the i-th area (q,) is in the range from 4 to 50 (4≤q _i ≤50), then the defect in this area has high compactness; if 50<q _i ≤200, then the defect in this area has moderate compactness; if q _i >200, then the defect in this area has low compactness;

- the number of defect areas with the smallest, average and greatest depths, in pieces (i _σ , pieces), which are determined according to the following rules: if the standard deviation of the brightness of a defect in the i-th area (a,) is within the range from the smallest standard deviation of the brightness of a defect from the identified standard deviations of the brightness of defects for all areas (σ _min ) to half the sum of the smallest standard deviation of the brightness of a defect from the identified standard deviations of the brightness of defects for all areas and the average standard deviation of the brightness of a defect from the identified standard deviations of the brightness of defects for all areas then the defect in the given area has the smallest depth; if σ _i is within the range of half the sum of the largest standard deviation of the defect brightness from the identified standard deviations of the defect brightness for all areas and the average standard deviation of the defect brightness from the identified standard deviations of the defect brightness for all areas to the greatest standard deviation of the defect brightness from the identified standard deviations of the defect brightness for all areas (σ _max ), then the defect in this area has the greatest depth; if then the defect in this area has an average depth.

The results of the work of the SNS and BRM (in particular, a map of barrel bore defects on digital images of the scan of its full cross-section, the number of areas for seven different defects, the total area of each defect, the number of areas for each defect that have high, moderate and low compactness, and the smallest, average and greatest depth) are output (displayed) to the PC monitor via the output unit of the results of non-destructive testing of the barrel bore of the SPV.

Fig. 5 and Fig. 6 show examples of displaying the results of non-destructive testing of the barrel bore of a rifled or smoothbore SPV on a PC monitor, obtained by an operator who does not have special training and/or qualifications, in any terrain conditions (laboratory, workshop, training ground, etc.).

One of the essential distinctive features of the patented invention, in comparison with known technical solutions of analogues and prototypes, is the use of the PAC (Fig. 1), having a certain composition and architecture (Fig. 2), with the help of which non-destructive testing of the bore of the SPV is directly carried out. Changing the PAC, as well as using another PAC, will not allow achieving the declared technical result.

The second essential distinctive feature of the invention being patented is the order (algorithm) of performing non-destructive testing of the SPV barrel bore, in particular the measures for its preparation and implementation. Changing the order (algorithm) of performing non-destructive testing of the SPV barrel bore, as well as using another order (algorithm), will not allow achieving the claimed technical result.

The third essential distinctive feature of the invention being patented is the use of the YOLO architecture neural network for obtaining a map of the SPV barrel bore defects on digital images of the scan of its full section, trained on the basis of data from 7 of the most common defects of the SPV barrel, determined within the framework of the conducted research works (carbon deposits, cavities, corrosion, fumes, chrome chips, rash and scratches). The use of a different order of detection, classification and determination of the boundaries of defects on digital images of the scan of the full section of the SPV barrel bore, as well as training the neural network on the basis of data of other defects of the SPV barrel bore or a different number of them, will not allow achieving the declared technical result.

The fourth essential distinctive feature of the invention being patented is the procedure for calculating the geometric and optical metrics of the SPV bore defects and the rules that make it possible to determine on the SPV bore defect map the number of areas for seven different defects, the total area of each defect, the number of areas for each defect that have high, moderate and low compactness, and the smallest, average and largest depth. Changing the calculation procedure, as well as using other rules, will lead to erroneous results.

The conducted patent research did not reveal any technical solutions that became publicly available in the world before the priority date of the claimed invention and characterized by the claimed set of features; therefore, it can be assumed that the said technical solution meets the patentability condition of “novelty”.

To apply the claimed invention, a simple hardware and software complex is used with universal, widely used and accessible equipment, tested in pilot industrial conditions, which corresponds to the patentability condition “industrial applicability”.

Claims (3)

1. A method for non-destructive testing of the bore of a small arms and cannon armament using a computer vision method, consisting of performing non-destructive testing of the bore by obtaining images of its surface with a camera with their subsequent processing by a computer vision system using a hardware and software complex, characterized in that any installed devices are first removed from the muzzle end of the bore and a device for input, output and rotation of a probe with a digital camera is tightly attached to it using nylon tape with its outer edge, a personal computer is turned on and software is loaded into which data are entered on the caliber and length of the bore in millimeters, the sector of shooting the bore surface by the camera for the specified caliber of small arms and cannon armament in degrees and the distance from the outer edge of the device for input, output and rotation of the probe with a digital camera to the muzzle end of the bore of the small arms and cannon armament in millimeters, then, after starting, the device for input, output and rotation of the probe with a digital the camera starts the movement of the probe with a digital camera to the cut of the breech of the barrel bore, after which, successively rotating 360 degrees, it moves back, shooting a full cut of the inner surface of the barrel bore and automatically transmitting the sequentially obtained digital images to the software, where a scan of the full cut of the barrel bore is formed from them, on which, using a convolutional neural network trained on a pre-formed database of carbon deposits, cavities, corrosion, burn meshes, chrome chips, rashes and scratches, detection, classification and determination of the boundaries of these defects are carried out, then in the block of their spatial marking, calculation of geometric and optical metrics, the perimeter of the defect in one area P is calculated according to the formula , where x, y are the coordinates of the point lying in the defect area, R _gr.def is the boundary of the defect area, the area of the defect in one area S according to the formula where R _def is the defect area, including the boundary, the compactness of the defect in one area q according to the formula average brightness value in one area of the defect m according to the formula where I is the image matrix, the standard deviation of the brightness of the defect in one area σ according to the formula total defect area S _total as a percentage of the total area of the full bore scan according to the formula where S _p is the area of the full bore scan, i is the number of defect areas, pieces, S _i is the area of the defect in the i-th area, and the number of defect areas with high, moderate and low compactness i _q is determined in pieces according to the rules: if the compactness of the defect in the i-th area q _i is in the range from 4 to 50, then the defect in this area has high compactness, if 50<q _i ≤200, then moderate compactness, if q _i >200, then low compactness; and the number of defect areas with the smallest, average and largest depths i _σ , in pieces according to the rules: if the standard deviation of the defect brightness in the i-th area σ _i is within the range from the smallest standard deviation of the defect brightness from the identified standard deviations of the defect brightness for all areas σ _min to half the sum of σ _min and the average standard deviation of the defect brightness from the identified standard deviations of the defect brightness for all areas σ _ср , then the defect in this area has the smallest depth, if σ _i is within the range from half the sum of the largest standard deviation of the defect brightness from the identified standard deviations of the defect brightness for all areas σ _max and σ _ср to σ _max , then - the greatest depth, if (σ _min +σ _ср )⋅0.5≤σ _i ≤(σ _max +σ _ср )⋅0.5, then - the average depth; then, through the output unit of the results of non-destructive testing of the bore of small arms and cannons, a map of the bore defects on digital images of the scan of its full cross-section, the number of areas for seven different defects, the total area of each defect, the number of areas for each defect having high, moderate and low compactness and the smallest, average and greatest depth are displayed on the monitor of the personal electronic computer. 2. The method according to paragraph 1, characterized in that, during preparation for non-destructive testing of the barrel bore, a ring is installed on the probe with a digital camera for centering it in the barrel bore with a diameter 0.2 mm smaller than the caliber of the small arms and cannon, which moves in the barrel bore together with the probe with a digital camera during non-destructive testing. 3. The method according to paragraph 1, characterized in that when conducting non-destructive testing of certain sections of the bore of a small arms and cannon weapon, data on the distances from the cut of the breech of the barrel to the beginning of each of the sections and their length in millimeters are entered into the software on a personal electronic computer before it is launched; when conducting non-destructive testing, a complete cut of the inner surface of the bore of the small arms and cannon weapon is photographed only in these sections sequentially when a probe with a digital camera moves from the breech to the muzzle.