Camera defibrillation method and system

Abstract

The application discloses a camera defibrillation method and system, which collects operation parameters of the camera, determines a defibrillation scheme of the camera based on the operation parameters, and controls a defibrillation component to execute the defibrillation scheme. The application can formulate different defibrillation schemes for different operation conditions, and realizes accurate control and correction of high-speed video stream image jitter.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer vision detection, and particularly relates to a camera de-shaking method and system. BACKGROUND

[0002] The packaging machine is a device commonly used in the field of SMT packaging, which can package chips in the process of high-speed operation. However, due to factors such as vibration and acceleration during device operation, the high-speed video stream images captured by the camera usually have serious shaking, thereby affecting the accuracy and reliability of subsequent image processing and analysis. Therefore, how to solve this problem has become one of the hotspots and difficulties in the application field of SMT packaging technology.

[0003] The existing image stabilization technology mainly includes optical anti-shake, digital anti-shake, mechanical stabilizer and software solution based on image processing algorithm. However, for the image shaking of the Jiufeng series fine packaging machine under high-speed video stream, the existing technology has the limitations and deficiencies of low precision, poor adaptability and high cost. SUMMARY

[0004] The present application provides a camera de-shaking method and system to solve the shaking of the camera end of the paster during the existing paster packaging process, which easily leads to inaccurate focusing, blurred and unclear chip images.

[0005] To solve the above technical problems, the technical solution provided by the present application is as follows:

[0006] A camera de-shaking method, comprising the following steps:

[0007] Collecting the running parameters of the camera, determining the de-shaking scheme of the camera based on the running parameters, and controlling the de-shaking component to execute the de-shaking scheme.

[0008] Preferably, determining the de-shaking scheme of the camera based on the running parameters comprises the following steps:

[0009] Using a fuzzy function to determine the membership set of the running parameters, and taking the pre-set de-shaking scheme corresponding to the membership set as the camera de-shaking scheme under the vibration acceleration of the camera.

[0010] Preferably, the running parameters include the vibration acceleration of the camera, the de-shaking component is a gas magnetic de-shaking device, and the gas magnetic de-shaking device is fixed on the camera; the gas magnetic de-shaking device comprises an air spring and a magnetic levitation vibration isolator arranged in the air spring; the membership set comprises a first set (0 < ω d < 5 Hz), a second set (5 ≤ ω d < 20 Hz) and a third set (20 < ω d < 50 Hz), ωd is the excitation frequency, and the values of the first set are smaller than the values of the second set, the values of the second set are smaller than the values of the third set;

[0011] When the excitation frequency belongs to the first set, the magnetic suspension vibration isolator is not started, and only the air spring is used for damping;

[0012] When the excitation frequency belongs to the second set, the magnetic suspension vibration isolator is started, and the corresponding current feedback is performed according to the value of the vibration acceleration;

[0013] When the excitation frequency belongs to the third set, the magnetic suspension vibration isolator is closed, and only the air spring is used for damping.

[0014] Preferably, the magnetic suspension vibration isolator comprises a plurality of winding electromagnets, an armature arranged between the plurality of winding electromagnets, and a bracket assembly for fixedly supporting the electromagnets and the armature.

[0015] Preferably, the bracket assembly comprises a bracket bottom plate, a first bracket side plate, a second bracket side plate, a bracket middle plate, and a bracket top plate, the air spring is arranged in a space between the bracket bottom plate and the bracket top plate, the first bracket side plate, the second bracket side plate, and the bracket middle plate are arranged in the air spring, the bracket bottom plate and the bracket top plate extend transversely, the first bracket side plate, the second bracket side plate, and the bracket middle plate all extend longitudinally, the first bracket side plate and the second bracket side plate are fixed to the bracket bottom plate, the bracket middle plate is fixed to the bracket top plate and arranged between the first bracket side plate and the second bracket side plate and away from the bracket top plate, and a transversely extending armature is fixed to a section of the bracket middle plate away from the bracket top plate, and winding electromagnets are fixed and arranged on the first bracket side plate and the second bracket side plate on both longitudinal sides of the armature.

[0016] Preferably, the method further comprises the following steps:

[0017] The image data collected by the camera is acquired, it is judged whether the image data is abnormal, if it is abnormal, the abnormal category is determined, and the corresponding abnormal processing scheme is performed on the image data to obtain denoised image data.

[0018] Preferably, the abnormal category includes inaccurate focusing, and the corresponding abnormal processing scheme of the inaccurate focusing is to denoise the image based on a dynamic damping algorithm based on Doppler effect, and the dynamic damping algorithm based on Doppler effect performs the following steps:

[0019] The original image data is demodulated to baseband to obtain a point target model of the original image data;

[0020] The distance compression processing is performed on the point target model to obtain a distance compressed signal of the original image data;

[0021] transforming data on each distance in the distance compressed signal to a range Doppler domain through azimuth FFT to obtain FFT transformed image data;

[0022] comparing the phase of the FFT transformed image data with a reference map, and determining a phase compensation value according to the comparison result;

[0023] phase compensating the FFT transformed image data based on the phase compensation value;

[0024] inverse FFT transforming the FFT transformed image data after phase compensation to obtain defibrated image data.

[0025] Preferably, the abnormality category includes image ghosting, and the abnormality processing scheme corresponding to the image ghosting is deep network sharpening processing, and the deep network sharpening processing performs the following steps:

[0026] constructing a ghosting annealing operator H of input image data and a pseudo-inverse matrix thereof;

[0027] constructing a ghosting removal diffusion network, and reconstructing a distribution space of input image data based on the ghosting removal diffusion network;

[0028] obtaining a content space of input image data, and obtaining a target annealing image based on the reconstructed distribution space and the content space

[0029] Preferably, the abnormality category includes image ghosting, and the abnormality processing scheme corresponding to the image ghosting is deep network sharpening processing, and the deep network sharpening processing performs the following steps:

[0030] A computer system comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the steps of the above method when executing the computer program.

[0031] The present application has the following beneficial effects:

[0032] 1、The camera defibrillation method and system in the present application, by collecting the running parameters of the camera, determining the defibrillation scheme of the camera based on the running parameters, and controlling the defibrillation component to execute the defibrillation scheme. The present application can formulate different defibrillation schemes for different running conditions, and realize accurate control and correction of high-speed video stream image jitter.

[0033] 2、In the preferred embodiment, the application uses gas magnetic suspension technology to reduce the vibration and jitter of the device during operation. By installing multiple precise magnetic suspension control systems on the device, the position, movement direction and speed of the device can be accurately controlled, thereby minimizing the vibration and jitter of the device. The application of this gas magnetic suspension technology not only improves the clarity and stability of the image, but also ensures the performance and reliability of the device.

[0034] 3、In the preferred embodiment, the application proposes a dynamic defibrillation algorithm based on the Doppler effect and combines an image sharpening model. The Doppler effect is used to compensate for the phase of the image to solve the problem of inaccurate focusing and image blur, and the deep network is used to make the image clearer.

[0035] 4、In the preferred embodiment, the application uses a bidirectional defibrillation structure. The gas magnetic defibrillation assembly is used at the object end to reduce the impact of mechanical vibration on the camera, and various image processing techniques are used at the viewing end to sharpen the blurred image caused by jitter, so that the obtained image is more stable and clear.

[0036] In addition to the purposes, features and advantages described above, the application has other purposes, features and advantages. The application will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application, and are incorporated in and constitute a part of this specification. The illustrations are provided to explain the application and are not intended to limit the application in any way. In the drawings:

[0038] Figure 1 is a partial structure diagram of the packaging machine of the embodiment of the application;

[0039] Figure 2 is a hardware circuit diagram of the packaging machine of the embodiment of the application;

[0040] Figure 3 is a structure diagram of the defibrillation assembly of the embodiment of the application;

[0041] Figure 4 is a flowchart of the defibrillation method of the embodiment of the application. DETAILED DESCRIPTION

[0042] The embodiments of the application will be described in detail below with reference to the accompanying drawings, but the application can be implemented in various different ways as limited and covered by the claims.

[0043] Embodiment one:

[0044] The application provides a camera defibrillation method, which is applied to a packaging machine U6, as shown in Figure 1As shown, the packaging machine U6 includes a camera assembly, a computing drive assembly, and a defibrillator assembly U4; the packaging machine U6 is used to package electronic components U16; the camera assembly is mounted on the packaging port of the packaging machine U6;

[0045] like Figure 2 As shown, the computing drive component includes a fuzzy controller U11, a CPU U12, a memory unit U17, an imaging drive module U19, an electromagnetic drive module U18, a video defibrillator module U21, an external interface module, and a bus U22; the CPU U12, the video defibrillator module U21, the external interface module, the imaging drive module U19, and the electromagnetic drive module U18 are connected via the bus U22; the external interface module U22 includes a monitoring interface and a sensor interface;

[0046] The camera assembly includes a CCD imaging chip U1, an imaging optical path U2, and an RGB camera lens U3; the imaging optical path U2 is equipped with a defibrillation component U4.

[0047] The CCD imaging chip U1, imaging optical path U2, RGB camera lens U3, and imaging drive module U19 are connected in sequence. The RGB camera lens U3 acquires the packaged image of electronic component U16 and transmits the packaged image to CCD imaging chip U1 through imaging optical path U2. Then, it is sent to CPU U12 through imaging drive module U19. CPU U12 sends it to memory unit U17 for storage and determines whether there is any abnormality in the packaged image. If there is an abnormality, the packaged image is sent to the visual defibrillation module for processing.

[0048] like Figure 3 As shown, the defibrillation assembly U4 includes an accelerometer U7, an air spring U8, a winding electromagnet U9, an armature U10, and a support assembly. The support assembly includes a support base plate, a first support side plate, a second support side plate, a support middle plate, and a support top plate. The air spring U8 is disposed in the space between the support base plate and the support top plate. The first support side plate, the second support side plate, and the support middle plate are disposed within the air spring U8. The support base plate and the support top plate extend laterally. The first support side plate, the second support side plate, and the support middle plate all extend longitudinally. The first support side plate and the second support side plate are fixed to the support base plate. The support middle plate is fixed to the support top plate and is disposed between the first support side plate and the second support side plate. A laterally extending armature U10 is fixed to a section of the armature U10 away from the support top plate. Winding electromagnets U9 are fixedly installed on the first support side plate and the second support side plate on both longitudinal sides of the armature U10. The accelerometer U7 is disposed on the outside of the top plate. The winding electromagnet U9 is electrically connected to the fuzzy control module U11.

[0049] The defibrillation assembly U4 in the application is a gas-magnetic defibrillation assembly, and defibrillation is realized through the synergistic effect of magnetic suspension vibration isolation and air spring. The air spring fills the air bag with compressed air to provide flexible support for the camera equipment, and can adapt to different working conditions under different loads by adjusting the inflation pressure. As a passive vibration isolator, it can effectively reduce the transmission of high-frequency dynamic vibration. The electromagnetic force of the magnetic suspension vibration isolator is directly transmitted to the cover plate on the air spring, thereby acting on the equipment. For low-frequency vibration, the low-frequency vibration of the equipment can be counteracted by controlling the size and direction of the electromagnetic force, and finally the isolation effect of the hybrid vibration isolator on high-frequency and low-frequency vibration is realized.

[0050] Assuming that the rotating patch machine generates exciting force Where ω d is the exciting frequency, F d is the amplitude of the exciting force, j represents the imaginary unit, and t represents time. Here, only the case when the hybrid vibration isolator is working normally (i.e. less than the maximum output electromagnetic force of the magnetic suspension vibration isolator) is considered, so the working principle of the gas-magnetic active-passive hybrid vibration isolator can be explained in three working conditions:

[0051] When the acceleration sensor input value is very small, i.e. ω d is very small, the vibration isolator is approximately in a static working state, at this time the corresponding bearing force is provided by the air spring, and the magnetic suspension vibration isolator does not work.

[0052] When the acceleration sensor input value is small, i.e. ω d is small, especially near the natural circular frequency of the air spring, the air spring amplifies the vibration, and the magnetic suspension vibration isolator works to reduce or even suppress the resonance of the air spring.

[0053] According to the accompanying Figure 3 , the magnetic suspension vibration isolator is composed of a bracket, a magnet, a differential coil, an armature, a displacement sensor and a force sensor. Since the magnetic force depends on the magnetic force generated by the current in the differential coil on the armature, and the displacement parameter obtained by the displacement sensor, the force derived by the electromagnetic according to the simplified electromagnetic circuit is as follows:

[0054]

[0055] When the parameters of the magnetic suspension vibration isolator are determined, k is a constant, k = μ0N 2 A / 4, where μ0(H / m) is the magnetic permeability of air, N is the number of turns of the coil, A is the magnetic pole area, x0 is the air gap, x is the air gap change distance (i.e. the moving distance of the armature), i1, i2 are the currents of the upper and lower coils respectively, p end is the pressure generated by the current state of the air chamber, and S is the contact area of the air and the contact surface.

[0056] According to equation (1), the equivalent stiffness K and damping coefficient C of the vibration isolator are derived as follows:

[0057] From the ideal thermodynamic equation, we have:

[0058]

[0059] where p0 is the initial state of the air chamber pressure, T0 is the initial state of the temperature (thermodynamic temperature), d is the initial air chamber height, and T2 is the temperature other than the initial state.

[0060]

[0061] At i1 = i 10 , i2 = i 20 , linearize F and C, and expand by Taylor formula:

[0062]

[0063] The linearized equivalent electromagnetic force, stiffness, and damping coefficient are shown in the above equations.

[0064] ③ When the acceleration sensor input value is large, i.e. ω d is large, the air spring has good vibration isolation effect, and the required output electromagnetic force of the magnetic suspension vibration isolator is very small, which is close to the non-working state.

[0065] (2) The fuzzy control module U11 of the air-magnetic vibration isolator is composed of a fuzzy reasoning engine, a fuzzification module, a defuzzification module, a rule base, and an input / output interface.

[0066] The working process of the fuzzy control module U11 based on the fuzzy control method is as follows:

[0067] Step 1: Input signal. The preprocessor acceleration signal and its rate detected by the acceleration sensor in the air-magnetic vibration isolator can be used as the input of the fuzzy controller.

[0068] Step 2: Fuzzification. Map the input acceleration sensor value a to the membership of the fuzzy set. Triangular or trapezoidal functions can be used to define the shape and range of the fuzzy set. According to expert experience, very small (NL) can be defined as 0-5 Hz, small (SL) as 5-20 Hz, and large (BL) as 20-50 Hz.

[0069] Step 3: Fuzzy rule base. According to the characteristics and design requirements of the system, create a series of fuzzy rules:

[0070] 1. If ω d belongs to NL, the magnetic vibration isolator does not work.

[0071] 2, if ω d belongs to SL, start the magnetic attraction device, and perform corresponding current feedback according to the value of acceleration a to achieve better air magnetic anti-shake effect.

[0072] 3, if ω d belongs to BL, turn off the magnetic attraction device, and the air spring achieves better anti-shake effect.

[0073] Step4: fuzzy reasoning. According to the input fuzzy set and fuzzy rule base, the fuzzy output membership is calculated using fuzzy reasoning method. This will get a fuzzy output set, which represents the possible control action.

[0074] Step5: defuzzification. The fuzzy output set is converted into specific control action. Defuzzification method can be used to obtain the specific output electromagnetic force size according to the defuzzification result.

[0075] Step6: feedback control. The current of the magnetic suspension isolator is the output of the fuzzy control module U11, which is used to adjust the current size of the electromagnet to achieve better defibrillation effect. All these language variables are defined as NB (negative big), NS (negative small), ZE (zero), PS (positive small) and PB (positive big).

[0076] Based on the structure of the above packaging machine, as Figure 4 shown, the camera defibrillation method in the application has the following specific steps:

[0077] When the packaging machine starts packaging, if vibration occurs, the acceleration sensor U7 collects the corresponding acceleration of the vibration and transmits the acceleration signal to the CPU U12 through the bus U22.

[0078] The CPU U12 determines whether to start the magnetic suspension defibrillation device and the corresponding current size of the magnetic suspension device based on the fuzzy control module U11 according to the acceleration signal size, and performs the end defibrillation.

[0079] Specifically, the fuzzy control module U11 includes three membership sets, namely the first set, the second set and the third set, and the value of the first set is less than the value of the second set, and the value of the second set is less than the value of the third set;

[0080] When the excitation frequency belongs to the first set, the air magnetic defibrillator is not started;

[0081] When the excitation frequency belongs to the second set, the magnetic attraction current size of the magnetic suspension isolator is determined based on the vibration acceleration, and the magnetic suspension isolator of the air magnetic defibrillator is controlled to execute the magnetic attraction current to realize air magnetic anti-shake;

[0082] When the excitation frequency belongs to the third set, the air spring of the gas magnetic defibrillator is started to perform isolation.

[0083] After starting the flow, the CPU U12 will perform anomaly detection and send an imaging control signal to the imaging driving unit U19 to control the RGB camera U3 to collect the image of the electronic component U16 to be adjusted on the CCD imaging chip U1 through the bus U22.

[0084] After processing by the imaging driving unit U19, the digital image will be transmitted to the CPU U12 through the bus U22, and also output to the memory unit U17 through the bus U22.

[0085] The CPU U12 continues to perform the anomaly detection task in the memory, judges the corresponding problems existing in the obtained electronic component RGB packaging image, and then performs Doppler effect defibrillation or deep sharpening network sharpening according to the corresponding problems.

[0086] Specifically, the anomaly detection task includes:

[0087] Edge detection is performed on the electronic component contour curve in the packaging image to obtain the region where the electronic component is located;

[0088] Due to the high-speed rotation of the packaging machine, the video stream obtained by the camera may not be in focus. Therefore, a refocusing network is used to compensate for the phase of the corresponding image through a Doppler effect dynamic defibrillation algorithm.

[0089] For the case of image blur, a DDNM deep sharpening network is used to sharpen the image.

[0090] For the case of image ghosting, first, the Doppler effect dynamic defibrillation method is used to compensate for the phase and adjust the image, and then the SRGAN deep sharpening network is used to better explore the information and features in the data.

[0091] The specific content of the Doppler effect dynamic defibrillation algorithm is as follows:

[0092] The Doppler effect dynamic defibrillation algorithm estimates the motion trajectory and speed of an object or camera by analyzing image sequences or sensor data.

[0093] Based on the principle of Doppler effect, the phase compensation amount of each pixel point is calculated to correct the jitter and blur in the image. By applying the calculated phase compensation amount to each pixel in the image, the phase adjustment is realized, so that the influence of moving objects or cameras is minimized.

[0094] Performing azimuth fast Fourier transform converts data into azimuth Doppler domain. The specific steps are as follows:

[0095] The raw image data is demodulated to baseband so that the range frequency is centered at zero. The demodulated point target model is

[0096]

[0097] A0is an arbitrary complex constant, Γ is the range time, η is the near range azimuth time, η c is the beam center offset time, T r is the transmitted pulse width; is the range envelope (rectangular window function), i.e.

[0098]

[0099] is the azimuth envelope (square function), i.e.

[0100]

[0101] f0is the image center frequency, K p is the range chirp rate, is the instantaneous slant range. The target azimuth time η is referenced to zero Doppler in the echo S0(Γ,η). For the case of multiple targets, a common absolute time η should be given, such as the data acquisition start time η = 0. According to the shift property of Fourier transform, the phase term

[0102] Range compression, let S0(f Γ ,η) be the range Fourier transform of S0(Γ,η), is the matched filter, then the range compression output is:

[0103]

[0104] where the compressed pulse network P r [Γ] is the window function, W r (f τ ) is the fast Fourier transform of W

[0105] Azimuth fast Fourier transform. At low grazing angles, the beam is pointed close to the zero Doppler direction, and the range equation can be approximated as a parabola:

[0106]

[0107] where V r is the effective relative velocity between the defibrillation device and the target, and R0is the nearest slant range.

[0108] From the above analysis (equation 10), the range compressed signal is:

[0109]

[0110] The azimuth phase modulation can be seen obviously from the second exponential term, since the phase is a function of η 2 , the signal has a linear frequency modulation characteristic, and the frequency modulation rate is (approximation under low squint angle):

[0111]

[0112] Subsequently, the data at each range is transformed to the range-doppler domain by azimuth FFT. For a given target, the first exponential term in S rc (Γ,η) is constant, so only the second exponential term needs to be considered in the signal derivation in the range-doppler domain. Using the principle of stationary phase (POSP), the time-frequency relationship in the azimuth direction is f η =-K a η, substituting η=-f η / K a into S rc (Γ,η), the signal of azimuth FFT is:

[0113]

[0114] W a (f η -f ηc ) is the frequency domain form of the pattern , and both are consistent in shape. The above formula contains two exponential terms, the first term is the target inherent phase information, which is independent of the image intensity. The second term is the frequency domain azimuth modulation with linear frequency modulation characteristic.

[0115] The phase of the image obtained by fast Fourier transform is compared with the reference pattern, and the phase compensation in the range-doppler domain can be defined as:

[0116]

[0117] The phase that needs to be compensated is:

[0118]

[0119] After the image bias phase compensation, the inverse fast Fourier transform is performed, and the formula is as follows:

[0120]

[0121] Where p a is the amplitude of the azimuth impulse response, and the output is the image after adjusting the phase according to the reference pattern.

[0122] Specifically, the specific content of the DDNM depth network clarification network is as follows:

[0123] First, the problem is mathematically modeled, define the residual annealing operator H and the pseudo-inverse matrix of the residual annealing operator H * , the operator H simulates the image residual problem existing in the camera end during the shooting process and the Gaussian blur. The residualization and blurring process can be simulated by the formula y=Hx, x represents a high-definition image without residual, and y represents a camera shooting picture. Thus, the input image y is image backstepping, and the target annealing image is backstepped The residualization is completed. Wherein:

[0124]

[0125] The formula (19) gives the constraint condition of the obtained residual removal image , and the expression form of the residual annealing operator H, in H, Δt represents a time range from the current frame to the future frame, V t+τ The image optical flow field description with time delay τ, σ v , σ t , the motion time standard deviation parameters respectively represent the influence of the object moving speed on the residual effect, and the effect of the influence effect decaying with time, q(x) is the data distribution function of the image.

[0126] To sum up, the bidirectional defibrillation structure is adopted, the gas-magnetic defibrillation assembly is used at the object end to reduce the influence of mechanical vibration on the camera, and various image processing techniques are used at the view end to clarify the blurred image generated due to the tremor, so that the obtained image is more stable and clear.

[0127] In addition, a dynamic defibrillation algorithm based on Doppler effect is proposed, and an image clarification model is combined, the Doppler effect is used to compensate the phase of the image to solve the problem of inaccurate focusing and image ghosting, and the deep network is used to make the image more clear.

[0128] The above only describes the preferred embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A camera defibrillation method characterized by, The method comprises the following steps: Collecting operation parameters of the camera, determining a defibrillation scheme of the camera based on the operation parameters, and controlling a defibrillation component to execute the defibrillation scheme; Determining a defibrillation scheme of the camera based on the operation parameters comprises the following steps: Using a fuzzy function to determine a membership set of the operation parameters, and taking a preset defibrillation scheme corresponding to the membership set as a camera defibrillation scheme under a vibration acceleration of the camera; The operation parameter includes a vibration acceleration of the camera, the defibrillation component is an air magnetic defibrillator fixed on the camera, the air magnetic defibrillator includes an air spring and a magnetic suspension vibration isolator arranged in the air spring, the membership set includes a first set, a second set and a third set, ω d is an exciting frequency, the value of the first set is less than the value of the second set, and the value of the second set is less than the value of the third set. When the excitation frequency belongs to the first set, the magnetic suspension vibration isolator is not started, and only the air spring is used for defibrillation; When the excitation frequency belongs to the second set, the magnetic suspension vibration isolator is started, and corresponding current feedback is performed according to the value of the vibration acceleration; When the excitation frequency belongs to the third set, the magnetic suspension vibration isolator is closed, and only the air spring is used for defibrillation; The first set is: 0 < ω d The second set is: 5 < ω d The third set is: 20 < ω d < 50 Hz; The magnetic suspension vibration isolator comprises a plurality of winding electromagnets, an armature arranged between the plurality of winding electromagnets, and a support assembly for fixedly supporting the electromagnets and the armature.

2. The camera defibrillation method of claim 1, wherein, The support assembly comprises a support bottom plate, a first support side plate, a second support side plate, a support middle plate, and a support top plate. The air spring is arranged in a space between the support bottom plate and the support top plate. The first support side plate, the second support side plate, and the support middle plate are arranged in the air spring. The support bottom plate and the support top plate extend transversely. The first support side plate, the second support side plate, and the support middle plate all extend longitudinally. The first support side plate and the second support side plate are fixed to the support bottom plate. The support middle plate is fixed to the support top plate and is arranged between the first support side plate and the second support side plate and away from the support top plate. A transversely extending armature is fixed to a section of the support middle plate away from the support top plate. Winding electromagnets are fixed to the first support side plate and the second support side plate on both longitudinal sides of the armature.

3. The camera defibrillation method of any of claims 1-2, wherein, The method further comprises the following steps: Obtaining image data collected by the camera, determining whether the image data is abnormal, determining an abnormality category if the image data is abnormal, and executing a corresponding abnormality processing scheme on the image data to obtain denoised image data.

4. The camera defibrillation method of claim 3, wherein, The abnormality category includes inaccurate focusing. The corresponding abnormality processing scheme for the inaccurate focusing is a dynamic defibrillation algorithm based on the Doppler effect for denoising the image. The dynamic defibrillation algorithm based on the Doppler effect executes the following steps: Demodulating the original image data to a baseband to obtain a point target model of the original image data; Performing distance compression processing on the point target model to obtain a distance compressed signal of the original image data; Converting data on each distance in the distance compressed signal to a distance Doppler domain through azimuth FFT to obtain FFT transformed image data; Comparing the phase of the FFT transformed image data with a reference phase and determining a phase compensation value according to the comparison result; Performing phase compensation on the FFT transformed image data based on the phase compensation value; Performing inverse FFT on the FFT transformed image data after phase compensation to obtain defibrillated image data.

5. The camera defibrillation method of claim 3, wherein, The abnormality category includes image residual image. The corresponding abnormality processing scheme for the image residual image is deep network sharpening processing. The deep network sharpening processing executes the following steps: Construct a residual annealing operator H of input image data and a pseudo-inverse matrix thereof; Construct a de-residual diffusion network, and reconstruct a distribution space of input image data based on the de-residual diffusion network; Obtaining a content space of input image data, obtaining a target annealed image based on the reconstructed distribution space and the content space 6. The camera defibrillation method of claim 3, wherein, The abnormal category includes image ghosting, and an abnormal processing scheme corresponding to the image ghosting is to first perform de-noising on the image based on a dynamic defibrillation algorithm of Doppler effect, and then perform deep network sharpening processing on the de-noised image; it is judged whether the image after the deep network sharpening processing exists ghosting abnormality, and if so, the image after the deep network sharpening processing is processed through an SRGAN deep sharpening network.

7. A computer system comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the method of any one of claims 1 to 6 when executing the computer program.

Images