A multi-directional vision measurement method based on a non-contact extensometer

Abstract

The present application belongs to the test technology of material performance test, and particularly relates to a multi-directional visual measurement method based on a non-contact extensometer. The present application specifically includes a regional feature recognition method based on a binocular visual recognition mode, and is used for displacement measurement of multi-directional deformation of a product to be measured. The non-contact multi-directional extensometer based on visual recognition in the present application converts the traditional contact measurement mode into a non-contact measurement mode by establishing a binocular visual data acquisition mode, avoids the problem of lateral force in the measurement process of the traditional extensometer, improves the measurement range of the extensometer, and compared with the traditional monocular visual detection mode, the binocular visual detection in the present application is more conducive to measuring the three-dimensional deformation data of the product after stretching deformation, avoids the problem that the image data of the product after stretching deformation is greatly distorted, and thus the displacement measurement of the deformation is inaccurate.

Description

Technical Field

[0001] This invention belongs to the field of material performance testing technology, with IPC classification number G01B21 / 32, and specifically relates to a multi-directional visual measurement method based on a non-contact extensometer. Background Technology

[0002] As an instrument for measuring the deformation and displacement of materials during tensile testing, extensometers typically require a combination of multiple sensors and fixtures in practical applications. However, these combinations often present limitations, such as the limited measurement directions of the extensometer, the inability to perform simultaneous multi-directional measurements, and the inherent inconvenience of measuring materials with low stiffness during conventional contact measurements. These issues restrict the measurement range of extensometers during use, hindering efficient measurement of various products, including thin films.

[0003] Patent CN202210102753 provides a processing method for an automatic visual extensometer to measure a specified plastic elongation strength. However, the method for detecting the tensile strength of a product by picking up speckle feature points described in this patent does not clearly specify the feature recognition standard and positioning standard for the speckles. Since the shapes of the speckles in the product are relatively similar, the deformation of some speckles is small, and the position of the speckles changes after deformation, the visual-based speckle recognition accuracy will be poor, ultimately leading to inaccurate measurement of the product's elongation strength.

[0004] Patent CN201611223757 provides a lateral force stabilization measuring device for high-temperature axial strain fatigue testing. The extensometer described in this patent is a contact extensometer. In order to improve the clamping strength between the extensometer and the product under test, a reverse adjustment stabilizer is installed on the extensometer. At the same time, in order to eliminate the tensile lateral force caused by the weight of the extensometer itself, a lateral pressure elimination mechanism is also installed. As a result, the overall structure of the extensometer is relatively complex, and the overall structure of the extensometer is heavy and not easy to operate.

[0005] Therefore, to address the problems existing in the actual measurement of existing extensometers, this invention provides a multi-directional visual measurement method based on a non-contact extensometer. Summary of the Invention

[0006] To address the aforementioned problems, this invention provides a multi-directional visual measurement method based on a non-contact extensometer. Specifically, it includes establishing a regional feature recognition method based on binocular visual recognition to perform displacement measurement of the multi-directional deformation of the product under test.

[0007] Specifically, the binocular vision recognition method detects the deformation and displacement of the product under test by establishing a visual difference. The visual difference measurement method is used to measure displacement data in a three-dimensional environment, avoiding the image distortion problem caused by traditional planar data sensors during the measurement process.

[0008] Preferably, the binocular vision recognition method first calibrates a designated location point of the product under test, and then establishes the detection coordinates of the product under test and performs feature recognition based on the calibrated location point.

[0009] Preferably, the regional feature recognition method includes global deformation recognition and local micro-deformation recognition.

[0010] Preferably, the global deformation identification involves rasterizing the surface of the product under test to form a raster image, and monitoring the displacement of feature points in the raster image to monitor the deformation of the product under test in a specified direction.

[0011] Specifically, the feature points in the grid image are cross-shaped feature points. By collecting the deformation length data of the cross-shaped feature points, the deformation direction of the product under test can be determined, and the displacement of the multi-directional deformation based on binocular vision recognition can be estimated.

[0012] Preferably, the local micro-deformation identification, based on the global deformation identification, identifies local deformations separately by monitoring the change in pixel depth within a unit area of ​​the raster image.

[0013] Preferably, the displacement measurement of the multi-directional deformation specifically includes lateral deformation displacement measurement and longitudinal deformation displacement measurement.

[0014] Preferably, the binocular vision recognition method employs a binocular camera and fixture collaborative control method.

[0015] Preferably, after the internal and external parameters of the binocular camera are calibrated, the coordinated control method dynamically adjusts the position of the fixture according to the real-time position of the product under test, thereby adjusting the product under test to the designated position of the binocular camera within the measurement range.

[0016] Preferably, the binocular vision recognition method collects global deformation data and local micro-deformation data of the product under test during the stretching process, and combines the collected data with the tensile stress to establish a dynamic stretching transformation curve.

[0017] Specifically, the collected data is used to calculate the plastic strain ratio and Poisson's distribution ratio, and combined with the material properties to obtain the corresponding tensile strain data.

[0018] Preferably, the tensile stress is controlled in real time according to the dynamic tensile transformation curve to ensure the uniformity of the measurement of the product under test.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0020] (1) The non-contact multi-directional extensometer based on visual recognition described in this invention establishes a data acquisition method based on binocular vision, which transforms the traditional contact measurement method into a non-contact measurement method. This avoids the problem of lateral force in the traditional extensometer measurement process and improves the measurement range of the extensometer. At the same time, compared with the traditional monocular vision detection method, the binocular vision detection described in this invention is more conducive to measuring the three-dimensional deformation data of the product after stretching and deformation, avoiding the problem of inaccurate displacement measurement of deformation caused by large distortion of the image data after stretching and deformation of the product.

[0021] (2) Based on (1), the present invention establishes a collection method based on cross-shaped feature points. The cross-shaped feature points are combined with a binocular vision detection device to intuitively determine the deformation direction of the product under test. The deformation length of the cross-shaped feature points is used to assist in estimating the deformation amount of the product under test, thereby improving the range of tensile measurement directions of the product under test and further improving the measurement accuracy of the product under test. Attached Figure Description

[0022] Figure 1 A flowchart for non-contact multi-directional extensometer identification using visual recognition. Detailed Implementation

[0023] Example 1:

[0024] The multi-directional visual measurement method based on a non-contact extensometer described in this embodiment specifically includes establishing a regional feature recognition method based on binocular visual recognition to perform displacement measurement of the multi-directional deformation of the product under test.

[0025] like Figure 1 As shown, the regional feature recognition method includes global deformation recognition and local micro-deformation recognition.

[0026] The global deformation identification method involves rasterizing the surface of the product under test to form a raster image, and monitoring the displacement of feature points in the raster image to monitor the deformation of the product under test in a specified direction.

[0027] The feature points in the grid image are extracted by forming a cross-shaped structure composed of two intersecting grid lines. This cross-shaped structure is used as a feature point for monitoring the deformation of the product under test. Compared with other forms of feature points such as circular structures of a specified radius and the spot structures of the product under test, the cross-shaped structure has the advantage of being more intuitive about the deformation size of the grid lines. By judging the degree of deformation of two intersecting grid lines, the displacement direction of the deformation of the product under test can be quickly determined, and deformation data in this direction can be collected for detection.

[0028] The binocular vision recognition method described above employs a collaborative control approach using a binocular camera and a fixture.

[0029] In the aforementioned non-contact measurement, the binocular vision recognition method is separated from the fixture of the product under test for detection. However, during the separation detection process, if the binocular vision recognition method and the fixture are controlled separately, the initial positioning of the product under test within the visual detection range may be inaccurate, requiring subsequent manual adjustment and recalibration. To further overcome this problem, this invention establishes a control method based on a binocular camera and fixture system. First, the initial position of the product under test is detected by the binocular camera, and the initial position detection data is transmitted to the fixture controller. The fixture is then adjusted a second time to improve the recognition accuracy under non-contact binocular vision recognition.

[0030] The aforementioned binocular vision recognition method collects global deformation data and local micro-deformation data of the product under test during the stretching process, and establishes a dynamic stretching transformation curve by combining the collected data with the tensile stress.

[0031] When the product under test is subjected to tensile testing, the actual tensile stress varies due to differences in the material and performance properties of the product. For products with relatively conventional performance properties, tensile measurement parameters can be predetermined during extensometer testing to avoid damage during the measurement process. However, for products with ambiguous performance properties or whose material composition is not fully disclosed, especially thin products, the actual tensile measurement parameters cannot be fixed. To address this issue, this invention first uses binocular visual recognition to measure the displacement of multi-directional deformation of the product under test. Based on the measurement data and tensile stress, a dynamic tensile transformation curve is established. The tensile stress under a specified deformation amount is analyzed based on the dynamic tensile transformation curve. When the deformation amount exceeds the rated deformation range of the product under test, the tensile stress is automatically adjusted to avoid tensile damage to the product.

Claims

1. A multi-directional visual measurement method based on a non-contact extensometer, characterized in that, This includes establishing a region feature recognition method based on binocular vision recognition, which is used to measure the displacement of multi-directional deformation of the product under test; The aforementioned regional feature recognition method includes global deformation recognition and local micro-deformation recognition; The global deformation identification process involves rasterizing the surface of the product under test to form a raster image, and monitoring the displacement of feature points in the raster image to monitor the deformation of the product under test in a specified direction. The feature points in the grid image are extracted by forming a cross-shaped structure composed of two intersecting grid lines. This cross-shaped structure is used as a feature point for monitoring the deformation of the product under test. Compared with the dot structure of a specified radius and the spot structure of the product under test, the cross-shaped structure has the advantage of being more intuitive about the deformation size of the grid lines. By judging the degree of deformation of the two intersecting grid lines, the displacement direction of the deformation of the product under test can be quickly determined, and deformation data in this direction can be collected for detection. The aforementioned local micro-deformation identification, based on global deformation identification, identifies local deformations separately by monitoring the change in pixel depth within a unit area of ​​the raster image. The displacement measurement of the multi-directional deformation specifically includes lateral deformation displacement measurement and longitudinal deformation displacement measurement; The binocular vision recognition method described above employs a collaborative control approach using a binocular camera and a fixture. In non-contact measurement, the binocular vision recognition method is separated from the fixture of the product under test for detection. A control method based on binocular camera and fixture system is established. First, the initial position of the product under test is detected by binocular camera, and the initial position detection data is transmitted to the fixture controller. The fixture is then adjusted to improve the recognition accuracy under non-contact binocular vision recognition. The aforementioned binocular vision recognition method collects global deformation data and local micro-deformation data of the product under test during the stretching process, and combines the collected data with the tensile stress to establish a dynamic stretching transformation curve; When the product under test is subjected to tensile measurement, the displacement of the multi-directional deformation of the product under test is first measured by binocular vision recognition. A dynamic tensile transformation curve is established based on the measurement data and tensile stress. The tensile stress under a specified deformation is analyzed based on the dynamic tensile transformation curve. When the deformation exceeds the rated deformation range of the product under test, the tensile stress is automatically adjusted to avoid tensile damage to the product.

Images