Tiny vibration vision measurement method and system capable of resisting environmental motion interference

Abstract

The invention provides a non-contact micro-vibration vision measurement method resistant to environmental motion interference. The method comprises the following steps: S1, shooting video data of a vibration target; s2, extracting a tiny vibration signal from the video; s3, performing spectrum characteristic analysis on the tiny vibration signal, and calculating a frequency invariance parameter; s4, filtering an environment interference motion signal by using a frequency invariance parameter, and reserving a target object vibration signal; s5, performing spectral analysis on the vibration signal of the target object to obtain a vibration frequency; wherein the step S3 comprises the steps that the tiny vibration signals are divided into N sections of signals, the peak frequency of each section of signal is obtained, the frequency invariance parameter is used for representing the distance between the adjacent peak frequencies, when the frequencies of the adjacent tiny vibration signals are kept unchanged and are not 0, the value of the corresponding frequency invariance parameter is 0, and otherwise, the frequency invariance parameter is larger than 0. According to the invention, the vision measurement efficiency and the calculation precision are greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of computer vision technology and relates to a non-contact visual measurement method and system for micro-vibrations with resistance to environmental motion interference. Background Technology

[0002] Minute vibrations are ubiquitous phenomena in nature, such as bridge swaying caused by wind, structural vibrations generated by the operation of machinery, and chest rise and fall caused by human breathing. These minute vibrations contain important information reflecting the essential properties of objects and can be further applied to fields such as bridge structural health monitoring, equipment fault detection, and vital sign monitoring. Therefore, conducting minute vibration measurement work has significant application value.

[0003] With the development of computer vision technology, compared with traditional contact and non-contact vibration measurement methods (such as piezoelectric sensors, accelerometers, optical vibration measurement, photoelectric position sensors, fiber optic grating sensors, etc.), visual measurement methods have the characteristics of non-contact, dense measurement across the entire field, low cost, and flexible operation, and are widely used in industries, medicine, aerospace, military and scientific research.

[0004] Visual measurement methods typically utilize video capture devices such as cameras, mobile phones, or smartphones to photograph the target object, obtaining video data of the vibrating object. Then, the target object's region is manually marked in the video, and the vibration signal is extracted and subjected to spectral analysis to ultimately obtain the vibration frequency information. However, existing visual measurement methods rely on manual selection of the target area, resulting in cumbersome measurement procedures and low efficiency. Furthermore, when conducting vibration measurements outdoors, the target object often blends into the background environment, making it difficult for existing visual measurement methods to distinguish between the target object's vibration signal and background interference signals (such as swaying vegetation, pedestrian movement, and vehicle movement), leading to reduced measurement accuracy.

[0005] Furthermore, existing visual vibration measurement methods are computationally complex and cannot overcome environmental motion interference. For example, Chinese patent CN201910826468.9, a non-contact visual measurement method for micro-vibrations, and Chinese patent CN202010943253.8, a multi-target micro-vibration frequency measurement method based on Euler's perspective, both rely on manual selection of the target area, and their algorithms are not suitable for measurement scenarios with background environmental motion interference. Summary of the Invention

[0006] The purpose of this invention is to provide a non-contact visual measurement method and system for micro-vibrations that is resistant to environmental motion interference, in order to solve the technical problems of inefficiency due to reliance on manual selection, as well as the complexity and inaccuracy of calculation in the existing micro-vibration visual measurement process.

[0007] To achieve the above-mentioned objectives, the present invention provides the following technical solution: This invention provides a non-contact visual measurement method for minute vibrations that is resistant to environmental motion interference, comprising the following steps: Step S1: Use video acquisition equipment to capture video data of the vibrating target; Step S2: Extract minute vibration signals from the acquired video; Step S3: Perform spectral feature analysis on the extracted minute vibration signal and calculate the frequency invariance parameter; Step S4: Use frequency invariance parameters to filter out environmental interference motion signals and retain the vibration signal of the target object; Step S5: Process and analyze the spectral characteristics of the vibration signal of the target object, and calculate the vibration frequency of the target object; Step S3 includes: dividing the minute vibration signal into N segments and obtaining the peak frequency of each segment. The frequency invariance parameter is used to characterize the distance between adjacent peak frequencies. When the frequencies of adjacent minute vibration signals remain fixed and are not 0, the value of the corresponding frequency invariance parameter is 0; otherwise, the frequency invariance parameter is greater than 0.

[0008] Furthermore, step S2 includes: Step S21, Color space conversion: Perform color space conversion on the video data acquired in step S1, converting it from RGB space to YIQ space, and extract the luminance component Y of each frame image to form new video data D0; Step S22, Micro-vibration signal extraction: Edge detection is performed on the first frame of video data D0 to obtain an edge detection result image M. The coordinate information of each edge pixel in M ​​is recorded and stored in set Z, and the total number of edge pixels is denoted as n. Image M is multiplied by each frame in video data D0 to obtain new video data D1. Each frame of D1 is subtracted from its first frame to obtain the difference video D2. According to the coordinate information in Z, the pixel values ​​at the coordinate positions of n edge pixels in each frame of D2 are extracted to form n time-dimensional pixel value sequences. The time-dimensional pixel value sequences are the micro-vibration signals. The edge detection can use the Canny operator.

[0009] Furthermore, step S3 includes: Step S31: Perform time-frequency analysis on the small vibration signals extracted in step S2, divide the small vibration signals into N segments, perform Fourier transform on each segment to obtain its spectrum, and record the peak frequencies in the spectra of the N segment signals to obtain the peak frequency sequence X=[f1 i f2 i ,...,f N i], where 1≤i≤n; Step S32: Construct a point coordinate (f) from two adjacent elements in sequence X. j i , f j+1 i , ), denoted as P j (where 1≤j≤N-1), a total of N-1 point coordinates; Step S33: Calculate the frequency invariance parameter β of the i-th minute vibration signal. i The formula is as follows: ; in, This represents two adjacent coordinate points P. k and P k+1 The Manhattan distance between them, when the frequency of the tiny vibration signal remains constant and is not zero, β i The value of β is 0, otherwise β i Greater than 0.

[0010] In step S31, the minute vibration signal is divided into N segments using a Hamming window function.

[0011] Furthermore, step S4 includes: Step S41: Calculate the frequency invariance parameter β of each minute vibration signal according to formula (1). i By filtering out signals whose frequency invariance parameter is greater than a preset threshold and retaining signals whose frequency invariance parameter is less than or equal to the preset threshold, the interference of background motion can be eliminated, and the vibration signal of the target object can be obtained.

[0012] Preferably, the preset threshold value range is [0, 0.01].

[0013] Furthermore, step S5 includes: The vibration signals of all target objects extracted in step S4 are summed and the mean is subtracted. Then, a Fourier transform is performed on the signal to obtain the spectrum of the target object's vibration signal. The frequency with the highest energy in the spectrum is the vibration frequency of the target object.

[0014] The present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the above-described method.

[0015] The present invention also provides a computer system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the memory implements the above-described method when executing the program.

[0016] This invention creatively provides a visual measurement method for minute vibrations suitable for outdoor scenarios, which can replace traditional measurement techniques; its beneficial effects include: 1. This invention is an improved visual measurement method for minute vibrations, which can realize automatic measurement without the need for manual marking and selection of target areas, thus greatly improving detection efficiency.

[0017] 2. This invention can automatically distinguish between the vibration signal of the target object and the motion signal of the background environment, thereby effectively filtering out the interference signal of the background environment motion, having the ability to resist background motion interference, and effectively detecting the vibration frequency information of the target object, thus improving the measurement accuracy. Attached Figure Description

[0018] Figure 1 This is a flowchart of the visual measurement method for micro-vibrations that resists environmental motion interference according to the present invention. Figure 2 This is a schematic diagram of the Y component, edge detection results, and small vibration signal waveform of the first frame image of a vibrating object provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the peak frequency sequence of vibration signals of the target object and the interfering object provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the vibration frequency detection results of a target object provided in an embodiment of the present invention. Detailed Implementation

[0019] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. These embodiments are intended to illustrate the present invention but are not intended to limit its scope.

[0020] Combination Figure 1 This invention proposes a visual measurement method for minute vibrations that resists environmental motion interference. The method mainly includes five steps: video acquisition, minute vibration signal extraction, spectral feature analysis, separation of target object vibration signal from environmental interference motion signal, and calculation of target object vibration frequency.

[0021] Step S1: Video capture; In this step, according to an embodiment of the present invention, video data of a vibrating object in a real scene is captured using the camera of a video acquisition device such as a mobile phone or camera.

[0022] For example, the frame rate is set to 59.95 frames per second, the image resolution to 1920×1080, the video acquisition time to 21 seconds, and the video format to MOV. To improve the measurement speed, the video image resolution is downsampled to 640×360 and saved as an AVI video.

[0023] Step S2: Extraction of minute vibration signals (combined with...) Figure 2 Extracting minute vibration signals from the video acquired in step S1, including: Step S21: Color Space Conversion. The video data acquired in step S1 is converted from RGB space to YIQ space, and the luminance component Y of each frame is extracted to form new video data D0.

[0024] Step S22: Extraction of minute vibration signals. Edge detection is performed on the first frame of video data D0 using the Canny operator in digital image processing, resulting in an edge detection image M. The coordinates of each edge pixel in M ​​are recorded and stored in set Z (denoted as n edges). Image M is multiplied by each frame of video data D0 to obtain new video data D1. Each frame of D1 is subtracted from its first frame to obtain the difference video D2. Based on the coordinate information in Z, pixel values ​​at n coordinate positions are extracted from each frame of D2, forming n time-dimensional pixel value sequences. These time-dimensional pixel value sequences represent the minute vibration signals.

[0025] Step S3: Spectral Feature Analysis. Perform spectral feature analysis on the extracted vibration signal to calculate the frequency invariance parameters. Combined with... Figure 3 ,include: Step S31: Perform time-frequency analysis on the small vibration signals extracted in step S2, that is, use the Hamming window function to divide the small vibration signals into N segments, and perform Fourier transform on each segment to obtain its spectrum. Record the peak frequencies in the spectra of the N segmented signals to obtain the peak frequency sequence [f1]. i f2 i ,...,f N i Let X be an integer. Where 1 ≤ i ≤ n.

[0026] Step S32: Construct a point coordinate (f) from two adjacent elements in sequence X. j i , f j+1 i , ), denoted as P j (where 1≤j≤N-1), there are a total of N-1 point coordinates.

[0027] Step S33: Calculate the frequency invariance parameter β of the i-th minute vibration signal. i The formula is as follows: ; in, This represents two adjacent coordinate points P. k and P k+1The Manhattan distance between them. Formula (1) is used to characterize the constancy of adjacent peak frequencies. When the frequency of a small vibration signal remains constant and is not zero, then β i The value of β is 0, otherwise β i Greater than 0.

[0028] Step S4: Separate the target object vibration signal from the environmental interference motion signal. Use frequency-invariant parameters to filter out environmental interference motion, retaining the target object vibration signal. More specifically, this includes: Step S41: Calculate the frequency invariance parameter β of each minute vibration signal according to formula (1). i Filter out β i Signals >0, retain β i A signal equal to 0 can eliminate interference from background motion and obtain the vibration signal of the target object. In addition, the decision threshold can be taken from the interval [0, 0.01] in addition to 0.

[0029] Step S5: Calculate the vibration frequency of the target object. (Combined with...) Figure 4 As shown, the spectral characteristics of the vibration signal of the target object are processed and analyzed to obtain the vibration frequency.

[0030] Step S51: Sum the vibration signals of all target objects extracted in step S4 and subtract the mean, then perform a Fourier transform on them to obtain the spectrum of the vibration signal of the target object; the frequency with the highest energy in the spectrum is the vibration frequency of the target object.

[0031] The present invention also provides a corresponding computer system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the memory executes the program to implement the steps in the above embodiments.

[0032] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A non-contact visual measurement method for minute vibrations resistant to environmental motion interference, characterized in that, Includes the following steps: Step S1: Use video acquisition equipment to capture video data of the vibrating target; Step S2: Extract minute vibration signals from the acquired video; Step S3: Perform spectral feature analysis on the extracted minute vibration signal and calculate the frequency invariance parameter; Step S4: Use frequency invariance parameters to filter out environmental interference motion signals and retain the vibration signal of the target object; Step S5: Process and analyze the spectral characteristics of the vibration signal of the target object, and calculate the vibration frequency of the target object; Step S3 includes: dividing the minute vibration signal into N segments and obtaining the peak frequency of each segment. The frequency invariance parameter is used to characterize the distance between adjacent peak frequencies. When the frequencies of adjacent minute vibration signals remain fixed and are not 0, the value of the corresponding frequency invariance parameter is 0; otherwise, the frequency invariance parameter is greater than 0.

2. The method as described in claim 1, characterized in that, Step S2 includes: Step S21, Color space conversion: Perform color space conversion on the video data acquired in step S1, converting it from RGB space to YIQ space, and extract the luminance component Y of each frame image to form new video data D0; Step S22, Micro-vibration signal extraction: Perform edge detection on the first frame of video data D0 to obtain edge detection result image M. Record the coordinate information of each edge pixel in M ​​and store it in set Z. The total number of edge pixels is n. Multiply the image M with each frame in video data D0 to obtain new video data D1. Subtract each frame of D1 from its first frame to obtain difference video D2. According to the coordinate information in Z, extract the pixel values ​​at the coordinate positions of n edge pixels in each frame of D2 to form n time-dimensional pixel value sequences. The time-dimensional pixel value sequences are the micro-vibration signals.

3. The method as described in claim 2, characterized in that, The edge detection uses the Canny operator.

4. The method as described in claim 2, characterized in that, Step S3 includes: Step S31: Perform time-frequency analysis on the small vibration signals extracted in step S2, divide the small vibration signals into N segments, perform Fourier transform on each segment to obtain its spectrum, and record the peak frequencies in the spectra of the N segment signals to obtain the peak frequency sequence X=[f1 i f2 i ,...,f N i ], where 1≤i≤n; Step S32: Construct a point coordinate (f) from two adjacent elements in sequence X. j i , f j+1 i , ), denoted as P j , where 1≤j≤N-1, and there are a total of N-1 point coordinates; Step S33: Calculate the frequency invariance parameter β of the i-th minute vibration signal. i The formula is as follows: ； in, This represents two adjacent coordinate points P. k and P k+1 The Manhattan distance between them, when the frequency of the tiny vibration signal remains constant and is not zero, β i The value of β is 0, otherwise β i Greater than 0.

5. The method as described in claim 4, characterized in that, Step S31 uses the Hamming window function to divide the minute vibration signal into N segments.

6. The method according to any one of claims 1-5, characterized in that, Step S4 includes: Step S41: Filter out signals whose frequency invariance parameter is greater than a preset threshold and retain signals whose frequency invariance parameter is less than or equal to the preset threshold. This will eliminate the interference of background motion and obtain the vibration signal of the target object.

7. The method as described in claim 6, characterized in that, The preset threshold value range is [0, 0.01].

8. The method according to any one of claims 1-5, characterized in that, Step S5 includes: The vibration signals of all target objects extracted in step S4 are summed and the mean is subtracted. Then, a Fourier transform is performed on the signal to obtain the spectrum of the target object's vibration signal. The frequency with the highest energy in the spectrum is the vibration frequency of the target object.

9. A computer storage medium storing a computer program that, when executed by a processor, implements the method as described in any one of claims 1-8.

10. A computer system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the memory executes the program, it implements the method as described in any one of claims 1-8.

Images