Image imaging method and image acquisition device based on anti-ambient light interference technology

Abstract

The present invention relates to the field of image processing technology, and in particular to an image imaging method and image acquisition device based on anti-ambient light interference technology. The method utilizes a photoelectric sensor to receive a mixed light signal reflected by a target, extracts the reflected light component containing pulse coding characteristics through the tunable optical filter, and generates an anti-interference feature map with a time dimension identifier; based on the time dimension identifier of the anti-interference feature map, identifies the effective reflection area in the target scene that meets the preset reflection characteristic threshold, and marks the transient interference area generated by direct ambient light; performs a differential operation on the light intensity data of the effective reflection area and the spectral characteristics of the transient interference area to reconstruct the intrinsic image of the target obscured by light. The present invention can improve the imaging clarity and target recognition reliability under harsh lighting conditions such as backlight and strong light.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an image imaging method based on anti-ambient light interference technology and an image acquisition device thereof. Background Art

[0002] In the field of outdoor image acquisition, ambient light interference (such as strong sunlight, backlight, and transient reflections) seriously affects image quality, leading to loss of target details and a decrease in signal-to-noise ratio, thus hindering reliable target recognition. Traditional anti-interference methods mainly rely on optical filtering, exposure adjustment, or post-processing algorithms, but they have significant limitations: fixed bandpass filters are difficult to adapt to the dynamically changing ambient light spectrum; long exposure or HDR technology is prone to introducing motion blur; and pure algorithmic denoising relies on a priori assumptions and fails under extreme lighting conditions. In recent years, active fill light combined with dynamic filtering has become a research hotspot, but existing technologies mostly use fixed-time fill light or wide-spectrum filtering, which cannot accurately separate the self-fill light from the ambient light components. In particular, signal aliasing still exists in scenes with strong light fluctuations. In addition, traditional devices lack closed-loop control mechanisms for ambient light adaptation, making it difficult to optimize imaging parameters in real time, resulting in insufficient robustness in dynamic scenes. Therefore, there is an urgent need for an imaging method and equipment that integrates time-frequency-space multi-dimensional anti-interference technology. By combining high-precision light source coding, dynamic spectrum filtering and time domain differentiation, stable imaging in complex light environments can be achieved to meet the high reliability requirements of outdoor intelligent sensing systems. Summary of the Invention

[0003] The present invention overcomes the deficiencies of the prior art and provides an image imaging method and image acquisition device based on an anti-ambient light interference technology.

[0004] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is:

[0005] The first aspect of the present invention discloses an image forming method based on the anti-ambient light interference technology, comprising the following steps:

[0006] S102: transmitting a fill light signal with a specific time code to the target scene through a pulse light source built into the image acquisition device, and simultaneously loading a dynamic bandpass parameter linked to the ambient light intensity onto the tunable optical filter;

[0007] S104: using a photoelectric sensor to receive a mixed light signal reflected by the target, extracting a reflected light component containing a pulse coding feature through the tunable optical filter, and generating an anti-interference feature map with a time dimension identifier;

[0008] S106: Based on the time dimension identifier of the anti-interference feature map, identifying the effective reflection area in the target scene that meets the preset reflection characteristic threshold, and marking the transient interference area caused by direct ambient light;

[0009] S108: performing a differential operation on the light intensity data of the effective reflection area and the spectral characteristics of the transient interference area to reconstruct an intrinsic image of the target obscured by light;

[0010] S110: Dynamically adjusting the encoding timing of the pulse light source and the bandpass response curve of the tunable optical filter according to the contrast index of the target intrinsic image and the ambient light fluctuation frequency.

[0011] Preferably, the step 102 is specifically:

[0012] The image acquisition device synchronously triggers the pulse light source to emit a pulse fill light signal with a preset time code. The pulse width of the pulse fill light signal matches the camera exposure time and forms a discrete fill light period within each exposure cycle. At the same time, the fill light timing mark is recorded;

[0013] The ambient light intensity is collected in real time through the light sensor, and the main frequency band of the ambient light is extracted based on the ambient light spectrum characteristics to generate dynamically updated interference light bandpass suppression parameters;

[0014] Calculating the degree of spectral overlap between the main frequency band of the ambient light and the center frequency band of the fill light signal, and dynamically adjusting the bandpass range of the tunable optical filter according to the spectral overlap, so that the passband center frequency of the tunable optical filter is locked to the fill light signal frequency band, and the bandwidth is adaptively contracted with the ambient light intensity to suppress interference;

[0015] The fill light timing mark is used to extract the sub-frame image corresponding to the fill light period from the acquired image. The time domain difference operation is performed based on the filtered frequency domain features to eliminate the constant component of the ambient light in the time domain and retain the dynamic component of the fill light signal modulation.

[0016] The spatial domain analysis of the sub-frame image after difference is performed to obtain the residual noise area. Mask matching is performed based on the spatial distribution template of the fill light signal to eliminate the low-frequency diffuse noise related to the ambient light and output the anti-interference target image.

[0017] Preferably, the step 104 is specifically:

[0018] The photoelectric sensor receives the mixed light signal reflected by the target, which contains the self-pulse fill light component and the ambient light interference component;

[0019] Inputting the mixed optical signal into a tunable optical filter, the tunable optical filter performs frequency domain filtering based on preloaded dynamic bandpass parameters to selectively pass a central frequency band containing pulse coding features and suppress a main frequency band of ambient light, and outputs a filtered reflected light signal;

[0020] Based on the filtered reflected light signal, using the fill light timing mark recorded by the pulse light source, extracting a subframe image sequence corresponding to the fill light period;

[0021] Performing a time domain difference operation on the subframe image sequence, eliminating the constant component of the ambient light in the time domain by calculating the pixel difference between adjacent subframes, and retaining the dynamic reflection component modulated by the pulse signal;

[0022] Based on the dynamic reflection component, an anti-interference feature map with a time dimension identifier is generated, and the feature map represents the reflection characteristic change of the target in a time axis encoding manner.

[0023] Preferably, a time domain difference operation is performed on the subframe image sequence to eliminate the constant component of the ambient light in the time domain by calculating the pixel difference between adjacent subframes, while retaining the dynamic reflection component modulated by the pulse signal, specifically:

[0024] Based on the timing mark of the pulse light source, the sub-frame image sequence is spatially aligned to eliminate pixel offset caused by target motion or camera shake, and the brightness baseline of each sub-frame is unified to eliminate the DC bias of ambient light;

[0025] According to the coding order of the fill light pulse, the difference values ​​of the corresponding pixels between adjacent subframes are calculated to generate a time domain gradient map, in which the constant component of the ambient light is offset by the difference operation, while the dynamic reflection component of the pulse modulation is retained as the gradient extreme value;

[0026] The time domain gradient image is spatially filtered, and the spatial distribution template of the pulse signal is used to match the effective reflection area to suppress the random noise caused by ambient light fluctuations. At the same time, the gradient extreme value synchronized with the pulse timing is amplified through nonlinear gain.

[0027] The enhanced gradient extreme value is reversely integrated according to the pulse coding timing to reconstruct the dynamic reflection intensity curve of the target during the fill light period, and the drift error accumulated during the integration process is eliminated to output a pure pulse modulated reflection signal.

[0028] Preferably, the step 106 is specifically:

[0029] Based on the time dimension identification of the anti-interference feature map, the temporal change rate of its reflection intensity is calculated pixel by pixel along the time axis to generate a time domain response intensity matrix;

[0030] Based on the time-domain response intensity matrix, pixels are divided into a set of candidate reflection regions with similar time-domain response patterns through cluster analysis, and a set of abnormal regions with discrete response patterns are marked;

[0031] For the candidate reflection area set, calculate the average reflection response intensity of each area at the rising and falling edges of the fill light pulse; when the average reflection response intensity is greater than the preset reflection characteristic threshold, determine it as a valid reflection area and record its area boundary coordinates;

[0032] Perform time-domain gradient analysis on the abnormal region set to extract the peak value of its reflection intensity mutation within a single exposure cycle; if the occurrence time of the mutation peak has no correlation with the pulse light source coding timing, and the peak width is less than the preset transient interference judgment threshold, it is marked as a transient interference pulse region;

[0033] The spatial proximity between the center coordinates of the transient interference pulse area and the boundary coordinates of the effective reflection area is calculated; if the distance between the two is less than the ambient light diffuse reflection radius, the transient interference pulse area is associated with the direct ambient light and marked as the direct ambient light transient interference area;

[0034] The boundary coordinates of the effective reflection area and the mask coordinates of the ambient light direct transient interference area are integrated to generate a region identification map with classification labels.

[0035] Preferably, the step 108 is specifically:

[0036] Based on the effective reflection area boundary coordinates and transient interference area mask coordinates of the area identification map, a pixel-level spatial mapping relationship is established to generate a dual-channel area positioning matrix with topological markings;

[0037] The dynamic reflection component intensity data of the effective reflection area is used as the amplitude parameter, and the main frequency band spectrum characteristics of the ambient light in the transient interference area are used as the phase parameter to construct a complex feature vector with spatial correlation.

[0038] In the complex domain, the complex eigenvectors of the effective reflection area and the complex eigenvectors of the transient interference area are conjugate-dot-multiplied. The coherence component of the ambient light interference is solved through the amplitude-phase separation mechanism to generate a decorrelation difference feature map.

[0039] Performing Hilbert transform on the decorrelation difference feature map to extract the instantaneous phase of its analytical signal, and reconstructing the continuous phase distribution of the target surface reflectivity by phase unwrapping method to generate a phase gradient field;

[0040] Performing a spatial convolution operation on the phase gradient field and the dynamic reflection component of the effective reflection area, suppressing the frequency domain aliasing noise in the convolution process by using Wiener filtering, and outputting the target intrinsic reflectivity basis function;

[0041] According to the mask coordinates of the transient interference area, Laplace smoothing interpolation is performed on the boundary area of ​​the target intrinsic reflectivity basis function that overlaps with the ambient light diffuse reflection radius to eliminate the edge diffraction effect and finally reconstruct the target intrinsic image that is obscured by light.

[0042] Preferably, the step 110 is specifically:

[0043] Calculate the local contrast distribution matrix from the target intrinsic image, and simultaneously obtain the real-time fluctuation spectrum of the ambient light sensor to extract its dominant fluctuation frequency and harmonic components;

[0044] The phase compensation coefficient of the pulse light source coding timing and the filter bandpass contraction factor are generated according to the mean gradient of the local contrast distribution matrix and the dominant fluctuation frequency of the ambient light;

[0045] Reconstructing the rising and falling edge timings of the pulse light source based on the phase compensation coefficient, so that the duty cycle of the fill light pulse is dynamically scaled inversely proportional to the ambient light fluctuation frequency, and inserting an anti-interference time window that is staggered with the ambient light fluctuation beat in each exposure cycle;

[0046] The passband slope of the tunable optical filter is adjusted by using the filter bandpass contraction factor, so that the center frequency of the passband slope of the tunable optical filter remains consistent with the spectrum of the fill light signal, while the passband width is adaptively compressed according to the intensity of the harmonic component of the ambient light, and the roll-off distortion at the passband edge is suppressed;

[0047] The adjusted pulse coding timing and filter response curve are input into the time domain verification module of the anti-interference characteristic map. By calculating the signal-to-noise ratio improvement in the effective reflection area, the phase compensation coefficient and the filter bandpass contraction factor are iteratively optimized until the gradient variance of the local contrast distribution matrix reaches the preset stability threshold.

[0048] A second aspect of the present invention discloses an image acquisition device, which is applied to any of the above-mentioned image imaging methods based on anti-ambient light interference technology, wherein the image acquisition device includes an optical component module, a signal processing module, a dynamic control module, and a mechanical structure module;

[0049] The optical component module includes:

[0050] A pulse light source unit, used for emitting a fill light signal with a specific time code, the wavelength range of which matches the tunable optical filter;

[0051] Tunable optical filter unit, used to dynamically adjust the bandpass parameters according to the ambient light intensity to suppress interference frequency bands;

[0052] A photoelectric sensor unit, configured to receive a mixed light signal reflected by a target and convert it into an electrical signal;

[0053] Ambient light sensor unit, used to monitor ambient light intensity and spectrum characteristics in real time;

[0054] The mechanical structure module includes:

[0055] The protective shell is made of light-shielding material and has a heat dissipation structure, with integrated optical components and circuits;

[0056] The pan / tilt mechanism is used to adjust the pitch and horizontal angle of the device, including an anti-shake motor and encoder;

[0057] Interface panel, providing power input, data communication and external trigger interface;

[0058] The mounting base is equipped with anti-vibration brackets and quick locking devices, suitable for outdoor fixed or mobile deployment;

[0059] The signal processing module includes:

[0060] Timing control unit, used to synchronize the encoding timing of the pulse light source with the camera exposure time;

[0061] A frequency domain filtering unit, used for dynamically bandpass filtering the mixed signal output by the photoelectric sensor;

[0062] A time domain difference unit is used to extract the dynamic reflection component of the fill light modulation and generate an anti-interference feature map;

[0063] An image reconstruction unit, configured to reconstruct a target intrinsic image based on a differential operation between an effective reflection area and an ambient light interference area;

[0064] The dynamic control module includes:

[0065] Pulse coding optimization unit, used to adjust the duty cycle and timing of the fill light signal according to the ambient light fluctuation frequency;

[0066] A filter parameter adjustment unit, used for dynamically optimizing the passband center frequency and bandwidth of the tunable optical filter;

[0067] A feedback calibration unit is used to iteratively optimize the light source and filter parameters based on the image contrast index.

[0068] Wherein, the pan-tilt mechanism includes:

[0069] Pitch motor, rotation range ±90°;

[0070] Horizontal rotary motor, speed adjustable from 0.1 to 30° / s;

[0071] Inertial measurement unit, used to compensate for vibration caused by wind vibration in real time.

[0072] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: the present invention effectively separates the target reflection signal and the ambient light interference through the coordinated processing of pulse coded fill light and dynamic filtering, accurately identifies and marks the effective target area, and reconstructs the image details obscured by strong light. At the same time, the system parameters are adaptively optimized through a closed-loop adjustment mechanism to improve the imaging clarity and target recognition reliability under harsh lighting conditions such as backlight and strong light. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, without paying any creative work, they can also obtain drawings of other embodiments based on these drawings.

[0074] Figure 1 This is a flow chart of the overall method of the image imaging method based on the anti-ambient light interference technology;

[0075] Figure 2 This is a partial method flow chart of the image imaging method based on the anti-ambient light interference technology;

[0076] Figure 3 This is a schematic structural diagram of the image acquisition device;

[0077] Figure 4 This is a schematic diagram of the structure of the mechanical structure module in this image acquisition device. DETAILED DESCRIPTION

[0078] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0079] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0080] like Figure 1 As shown, the first aspect of the present invention discloses an image imaging method based on anti-ambient light interference technology, comprising the following steps:

[0081] S102: transmitting a fill light signal with a specific time code to the target scene through a pulse light source built into the image acquisition device, and simultaneously loading a dynamic bandpass parameter linked to the ambient light intensity onto the tunable optical filter;

[0082] S104: using a photoelectric sensor to receive a mixed light signal reflected by the target, extracting a reflected light component containing a pulse coding feature through the tunable optical filter, and generating an anti-interference feature map with a time dimension identifier;

[0083] S106: Based on the time dimension identifier of the anti-interference feature map, identifying the effective reflection area in the target scene that meets the preset reflection characteristic threshold, and marking the transient interference area caused by direct ambient light;

[0084] S108: performing a differential operation on the light intensity data of the effective reflection area and the spectral characteristics of the transient interference area to reconstruct an intrinsic image of the target obscured by light;

[0085] S110: Dynamically adjusting the encoding timing of the pulse light source and the bandpass response curve of the tunable optical filter according to the contrast index of the target intrinsic image and the ambient light fluctuation frequency.

[0086] It should be noted that this invention solves the technical problem of image quality degradation and target feature obscuration caused by ambient light interference in image acquisition equipment under complex outdoor lighting conditions. Through the coordinated processing of pulse-coded fill light and dynamic filtering, the target reflection signal is effectively separated from ambient light interference, the effective target area is accurately identified and marked, and image details obscured by strong light are reconstructed. At the same time, a closed-loop adjustment mechanism adaptively optimizes system parameters, improving imaging clarity and target recognition reliability in harsh lighting conditions such as backlight and strong light.

[0087] Preferably, the 102, as Figure 2 As shown, specifically:

[0088] S202: The image acquisition device synchronously triggers the pulse light source to emit a pulse fill light signal with a preset time code. The pulse width of the pulse fill light signal matches the camera exposure time, and forms a discrete fill light period in each exposure cycle. At the same time, a fill light timing mark is recorded.

[0089] S204: collecting ambient light intensity in real time through a light sensor, extracting the main frequency band of the ambient light based on the ambient light spectrum characteristics, and generating a dynamically updated interference light bandpass suppression parameter;

[0090] S206: Calculating the spectral overlap between the main frequency band of the ambient light and the center frequency band of the fill light signal, and dynamically adjusting the passband range of the tunable optical filter according to the spectral overlap, so that the passband center frequency of the tunable optical filter is locked to the fill light signal frequency band, and the bandwidth is adaptively contracted with the ambient light intensity to suppress interference;

[0091] S208: extracting the subframe image corresponding to the fill light period from the captured image using the fill light timing mark, performing a time domain difference operation based on the filtered frequency domain features, eliminating the constant component of the ambient light in the time domain, and retaining the dynamic component of the fill light signal modulation;

[0092] S210: Performing spatial domain analysis on the sub-frame image after difference to obtain the residual noise area, performing mask matching based on the spatial distribution template of the fill light signal, eliminating the low-frequency diffuse reflection noise related to the ambient light, and outputting the anti-interference target image.

[0093] It should be noted that, first, spatial domain analysis is performed on the subframe images after differential processing to detect residual noise areas in the image. Then, a pre-calibrated spatial distribution template of the fill light signal (e.g., a known fill light illumination range) is matched to the image to generate a corresponding effective signal area mask. This mask is then used to filter out low-frequency noise associated with diffuse reflections of ambient light (e.g., background interference caused by scattered sunlight) in the image, retaining the target reflection areas that match the fill light signal. Finally, edge optimization and contrast enhancement are performed on the processed image to output a clear, interference-resistant target image. This method effectively separates the fill light signal from ambient light noise, improving image quality.

[0094] In one embodiment of the present invention, in an intelligent highway monitoring scenario, an image acquisition device (such as a traffic violation capture camera) is installed in a road section with strong backlighting. During implementation, the camera synchronously triggers an 850nm near-infrared pulse light source, emitting a square wave fill light signal with a 20% duty cycle at a 10ms period. The rising edge of the pulse is strictly aligned with the exposure timing of the CMOS sensor. If the ambient light sensor detects real-time fluctuations in sunlight intensity in the 700-1000nm band and identifies a main peak of sunlight interference near 900nm, it dynamically adjusts the passband of the tunable optical filter to 840-860nm and compresses the bandwidth to 5nm to suppress interference. Within each pulse cycle, the camera extracts the subframe corresponding to the fill light period, eliminates the constant sunlight component by differentially determining adjacent subframes, and then performs spatial domain matched filtering based on a preset fill light spot spatial distribution template (an elliptical illumination area with a diameter of 2m). The final output is a high-definition license plate image that eliminates sunlight glare.

[0095] Overall, the present invention uses synchronous coded pulse fill light and dynamic tunable filtering, combined with time-frequency domain joint processing, to effectively separate the fill light signal from the ambient light interference, improve the signal-to-noise ratio of the target image in strong light or backlight scenes, adaptively suppress environmental spectral interference, accurately extract the fill light modulation signal, and further eliminate residual noise through time domain difference and spatial domain matching, ultimately outputting a high-contrast, low-interference target image, and enhancing the imaging robustness under complex lighting conditions.

[0096] Preferably, the step 104 is specifically:

[0097] The photoelectric sensor receives the mixed light signal reflected by the target, which contains the self-pulse fill light component and the ambient light interference component;

[0098] Inputting the mixed optical signal into a tunable optical filter, the tunable optical filter performs frequency domain filtering based on preloaded dynamic bandpass parameters to selectively pass a central frequency band containing pulse coding features and suppress a main frequency band of ambient light, and outputs a filtered reflected light signal;

[0099] Based on the filtered reflected light signal, using the fill light timing mark recorded by the pulse light source, extracting a subframe image sequence corresponding to the fill light period;

[0100] Performing a time domain difference operation on the subframe image sequence, eliminating the constant component of the ambient light in the time domain by calculating the pixel difference between adjacent subframes, and retaining the dynamic reflection component modulated by the pulse signal;

[0101] Based on the dynamic reflection component, an anti-interference feature map with a time dimension identifier is generated, and the feature map represents the reflection characteristic change of the target in a time axis encoding manner.

[0102] It should be noted that the extracted dynamic reflection components are arranged according to the timing of the pulse fill light to form time series data. Then, the reflection intensity changes of each pixel along the time axis are encoded to generate a time dimension identifier. Normalization is used to eliminate brightness differences between different time periods, preserving the temporal characteristics of the target's reflection characteristics. Finally, the processed data is integrated into a two-dimensional feature map, where the horizontal axis represents time information and the vertical axis represents the reflection intensity changes. This creates an anti-interference feature map that clearly characterizes the target's dynamic reflection characteristics.

[0103] Preferably, a time domain difference operation is performed on the subframe image sequence to eliminate the constant component of the ambient light in the time domain by calculating the pixel difference between adjacent subframes, while retaining the dynamic reflection component modulated by the pulse signal, specifically:

[0104] Based on the timing mark of the pulse light source, the sub-frame image sequence is spatially aligned to eliminate pixel offset caused by target motion or camera shake, and the brightness baseline of each sub-frame is unified to eliminate the DC bias of ambient light;

[0105] According to the coding order of the fill light pulse, the difference values ​​of the corresponding pixels between adjacent subframes are calculated to generate a time domain gradient map, in which the constant component of the ambient light is offset by the difference operation, while the dynamic reflection component of the pulse modulation is retained as the gradient extreme value;

[0106] It should be noted that the continuously acquired multiple frames are arranged in chronological order according to the encoding time sequence of the pulse light source. The brightness values ​​of pixels at the same position in two adjacent frames are then subtracted to produce a differential image for each pair of adjacent frames. Since ambient light interference remains essentially unchanged between adjacent frames, it cancels out after the differential operation. However, the target reflection signal modulated by the pulsed light will form distinct areas of sudden brightness changes in the differential image due to the changes in the pulse on-off state. Finally, all the differential results are superimposed to generate a time-domain gradient map, where areas corresponding to the pulse signal exhibit high gradient values, while areas with background interference maintain low gradient values.

[0107] The time domain gradient image is spatially filtered, and the spatial distribution template of the pulse signal is used to match the effective reflection area to suppress the random noise caused by ambient light fluctuations. At the same time, the gradient extreme value synchronized with the pulse timing is amplified through nonlinear gain.

[0108] The enhanced gradient extreme value is reversely integrated according to the pulse coding timing to reconstruct the dynamic reflection intensity curve of the target during the fill light period, and the drift error accumulated during the integration process is eliminated to output a pure pulse modulated reflection signal.

[0109] It is important to note that the time-domain gradient map is compared with a pre-calibrated pulse spot spatial distribution template. Template matching is used to screen out valid reflection areas while filtering out mismatched random noise. Within these valid areas, the gradient extrema that are synchronized with the pulse timing are nonlinearly amplified to enhance signal characteristics. These amplified gradient values ​​are then reversely accumulated according to the pulse encoding time sequence to reconstruct the complete target reflection intensity curve during the fill light cycle. Finally, baseline correction is performed to eliminate drift errors that may have been introduced during the integration process, resulting in a purified target reflection signal.

[0110] For example, in a nighttime highway monitoring scenario (continuing the aforementioned backlight capture system), the image acquisition device receives the mixed light signal reflected by the vehicle license plate through a photoelectric sensor (including 940nm pulse fill light and 589nm street light interference), and the tunable optical filter dynamically locks the passband to 940±2nm to suppress ambient light; based on the pulse timing mark (period 5ms / duty cycle 30%), the aligned fill light sub-frame sequence is extracted, and pixel differential is performed on adjacent sub-frames (Δt=1ms) to eliminate constant interference. Then, the effective area is matched through an elliptical fill light template and the gradient extreme value is amplified. Finally, the license plate reflection curve is reconstructed by reverse integration, and a time-coded anti-interference feature map is output to achieve clear imaging of the license plate under strong street light conditions.

[0111] In general, this method uses a tunable optical filter to dynamically suppress interference in the main frequency band of ambient light, eliminates the constant component of ambient light through time-domain differential operation, accurately extracts the dynamic reflection characteristics of pulse modulation, and reconstructs the pure target reflection signal with the help of spatial domain alignment and noise suppression technology. Finally, it generates an anti-interference feature map, thereby improving the accuracy and stability of target detection in complex light environments.

[0112] Preferably, the step 106 is specifically:

[0113] Based on the time dimension identification of the anti-interference feature map, the temporal change rate of its reflection intensity is calculated pixel by pixel along the time axis to generate a time domain response intensity matrix;

[0114] Based on the time-domain response intensity matrix, pixels are divided into a set of candidate reflection regions with similar time-domain response patterns through cluster analysis, and a set of abnormal regions with discrete response patterns are marked;

[0115] It should be noted that for each pixel in the time-domain response intensity matrix, its reflection intensity variation characteristics along the time axis are extracted. A clustering algorithm (such as K-means) is then used to group pixels based on these temporal characteristics. Pixels with similar variation patterns are classified as candidate reflection regions (e.g., regions that change synchronously with the pulse). Pixels whose temporal characteristics significantly deviate from the primary pattern are labeled as abnormal regions (e.g., interference regions with sudden changes or random fluctuations).

[0116] For the candidate reflection area set, calculate the average reflection response intensity of each area at the rising and falling edges of the fill light pulse; when the average reflection response intensity is greater than the preset reflection characteristic threshold, determine it as a valid reflection area and record its area boundary coordinates;

[0117] Perform time-domain gradient analysis on the abnormal region set to extract the peak value of its reflection intensity mutation within a single exposure cycle; if the occurrence time of the mutation peak has no correlation with the pulse light source coding timing, and the peak width is less than the preset transient interference judgment threshold, it is marked as a transient interference pulse region;

[0118] The spatial proximity between the center coordinates of the transient interference pulse area and the boundary coordinates of the effective reflection area is calculated; if the distance between the two is less than the ambient light diffuse reflection radius, the transient interference pulse area is associated with the direct ambient light and marked as the direct ambient light transient interference area;

[0119] The boundary coordinates of the effective reflection area and the mask coordinates of the ambient light direct transient interference area are integrated to generate a region identification map with classification labels.

[0120] Taking the evening highway scenario as an example, in an environment with both sunset light interference (main peak wavelength 600nm) and transient interference from vehicle LED headlights, the rate of change of the license plate area's reflection intensity is calculated pixel by pixel based on the time-domain encoding of the anti-interference feature map, generating a time-domain response intensity matrix. K-means clustering (number of clusters = 3) is then used to partition the pixels into candidate regions with synchronous response pulses (e.g., the license plate character area) and discrete anomaly regions (e.g., sun glare). The mean reflection intensity of the candidate regions on the rising and falling edges of the pulses is calculated (e.g., the mean for the character area is >450 grayscale). Regions exceeding a threshold (preset 400 grayscale) are marked as valid reflection regions, and their boundary coordinates are recorded. For anomaly regions, the peak value of sudden changes within a single cycle (e.g., LED headlight interference pulse width 1.2ms) is extracted. If there is no synchronization with the pulse sequence and the pulse width is <2ms (preset transient threshold), it is marked as a transient interference pulse region. The distance between the center of the transient interference area and the effective area boundary is calculated (for example, the distance between sunlight glare and the license plate boundary is 0.8m). If it is less than the ambient light diffuse reflection radius (preset 1.5m), it is associated with direct transient interference. Finally, the license plate effective area boundary (green outline) and the transient interference mask (red block) are combined to generate a region identification map with classification labels for subsequent image reconstruction modules.

[0121] Overall, this method effectively distinguishes between real target reflections and ambient light interference based on temporal dimension characteristics, accurately calibrates the effective target area through dynamic threshold judgment and spatial correlation analysis, and intelligently identifies and marks various transient interference areas. Finally, it outputs a regional distribution map with clear classification identification, thereby improving the accuracy of target detection and environmental anti-interference ability.

[0122] Preferably, the step 108 is specifically:

[0123] Based on the effective reflection area boundary coordinates and transient interference area mask coordinates of the area identification map, a pixel-level spatial mapping relationship is established to generate a dual-channel area positioning matrix with topological markings;

[0124] The dynamic reflection component intensity data of the effective reflection area is used as the amplitude parameter, and the main frequency band spectrum characteristics of the ambient light in the transient interference area are used as the phase parameter to construct a complex feature vector with spatial correlation.

[0125] It should be noted that a precise correspondence between pixel positions is established based on the boundaries of the effective reflection area and the coordinates of the transient interference area marked in the area identification map. A two-channel matrix is ​​then created, with one channel recording the topological structure of the effective area and the other channel marking the distribution of the interference area. Dynamic light intensity data from the effective reflection area is extracted as amplitude information, while the main frequency band characteristics of the ambient light in the interference area are obtained as phase information. The amplitude and phase information are then combined based on spatial location to construct a complex feature vector.

[0126] In the complex domain, the complex eigenvectors of the effective reflection area and the complex eigenvectors of the transient interference area are conjugate-dot-multiplied. The coherence component of the ambient light interference is solved through the amplitude-phase separation mechanism to generate a decorrelation difference feature map.

[0127] It should be noted that the complex eigenvectors of the effective reflection area and the transient interference area are conjugate-dot-multiplied in the complex domain, separating the amplitude and phase components using the complex multiplication rule. The result is then subjected to amplitude-phase decomposition to extract the coherent component of the ambient light interference. This coherent component is then subtracted from the mixed signal to eliminate the coupling effect between the interference signal and the target reflection signal. Finally, the processed amplitude and phase information are recombined to generate a differential feature map that has been stripped of the coherent interference from the ambient light.

[0128] Performing Hilbert transform on the decorrelation difference feature map to extract the instantaneous phase of its analytical signal, and reconstructing the continuous phase distribution of the target surface reflectivity by phase unwrapping method to generate a phase gradient field;

[0129] It should be noted that the decorrelated differential feature map is Hilbert transformed (i.e., constructing an analytical signal through a 90° phase shift) to obtain its analytical signal, thereby extracting the instantaneous phase value of each pixel. Then, using phase unwrapping (detecting and compensating for phase jumps that are integer multiples of 2π by comparing the phase differences between adjacent pixels), the truncated phase values ​​are expanded into a continuous phase distribution. Based on this unwrapped continuous phase data, the phase change gradient at each location on the target surface is calculated, thereby generating a phase gradient field that reflects the target's true topography.

[0130] Performing a spatial convolution operation on the phase gradient field and the dynamic reflection component of the effective reflection area, suppressing the frequency domain aliasing noise in the convolution process by using Wiener filtering, and outputting the target intrinsic reflectivity basis function;

[0131] According to the mask coordinates of the transient interference area, Laplace smoothing interpolation is performed on the boundary area of ​​the target intrinsic reflectivity basis function that overlaps with the ambient light diffuse reflection radius to eliminate the edge diffraction effect and finally reconstruct the target intrinsic image that is obscured by light.

[0132] Similarly, in the highway sunset interference scenario, for license plate areas obscured by direct sunlight, a dual-channel positioning matrix is ​​constructed based on the license plate valid area boundary (pixel coordinate set P) and the transient interference mask (coordinate set Q) in the regional identification map. The dynamic reflected light intensity of the license plate area is used as the amplitude, and the 589nm spectral phase of the interference area is used as the parameter to form a complex eigenvector. A conjugate dot product is performed on the P / Q vector in the complex domain to resolve the coherent component of the sunlight interference (for example, separating 60% of the coherent energy), generating a decorrelation difference map. A Hilbert transform is performed on the difference map to extract the instantaneous phase. Phase unwrapping is then used to reconstruct the continuous phase distribution of the license plate surface (with a resolution of 0.1 radians), generating a phase gradient field. The phase gradient field is spatially convolved with the dynamic reflected light component of the license plate (convolution kernel 3×3). A Wiener filter (with a signal-to-noise ratio prior of 40dB) is used to suppress aliasing in the frequency domain, outputting the license plate intrinsic reflectance basis function. According to the coordinates of the interference mask Q (0.8m from the license plate boundary), Laplace smoothing interpolation is performed on the sunlight diffusion overlapping area (within a radius of 1.5m) in the reflectivity function to eliminate edge diffraction artifacts and finally reconstruct the complete license plate image obscured by the sunset light.

[0133] In general, the present invention effectively separates the target reflection signal from the ambient light interference based on complex operations related to spatial domain, reconstructs the complete surface features of the obscured target, eliminates the diffraction effect through intelligent edge repair, and finally outputs a high-fidelity target intrinsic image, thereby improving the integrity and detail restoration capabilities of image reconstruction in complex light environments.

[0134] Preferably, the step 110 is specifically:

[0135] Calculate the local contrast distribution matrix from the target intrinsic image, and simultaneously obtain the real-time fluctuation spectrum of the ambient light sensor to extract its dominant fluctuation frequency and harmonic components;

[0136] It should be noted that the target intrinsic image is divided into blocks, the local contrast value of each image block is calculated, and the contrast distribution matrix is ​​generated.

[0137] The phase compensation coefficient of the pulse light source coding timing and the filter bandpass contraction factor are generated according to the mean gradient of the local contrast distribution matrix and the dominant fluctuation frequency of the ambient light;

[0138] Reconstructing the rising and falling edge timings of the pulse light source based on the phase compensation coefficient, so that the duty cycle of the fill light pulse is dynamically scaled inversely proportional to the ambient light fluctuation frequency, and inserting an anti-interference time window that is staggered with the ambient light fluctuation beat in each exposure cycle;

[0139] The passband slope of the tunable optical filter is adjusted by using the filter bandpass contraction factor, so that the center frequency of the passband slope of the tunable optical filter remains consistent with the spectrum of the fill light signal, while the passband width is adaptively compressed according to the intensity of the harmonic component of the ambient light, and the roll-off distortion at the passband edge is suppressed;

[0140] The adjusted pulse coding timing and filter response curve are input into the time domain verification module of the anti-interference characteristic map. By calculating the signal-to-noise ratio improvement in the effective reflection area, the phase compensation coefficient and the filter bandpass contraction factor are iteratively optimized until the gradient variance of the local contrast distribution matrix reaches the preset stability threshold.

[0141] For example, in a highway section with multiple tunnels, where frequent vehicles entering and exiting tunnels cause significant ambient light fluctuations, the system extracts the local contrast distribution matrix (e.g., an average gradient value of 85 in the character area) from the reconstructed license plate intrinsic image. Simultaneously, the ambient light sensor detects the interference from the fluorescent light at a main frequency of 100Hz and its second harmonic of 200Hz, generating a phase compensation coefficient of 0.25. The 940nm pulse duty cycle is then dynamically reduced from 30% to 15% (inversely proportional to 100Hz), and a 0.5ms anti-interference window is inserted every 5ms (to offset the peak phase of the fluorescent light). A shrinkage factor of 0.6 is used to compress the filter passband: the center frequency remains at 940nm, the bandwidth is compressed from 4nm to 2.4nm, and the passband edge roll-off rate is increased to 60dB / octet. The new parameters are then input into the verification module to calculate the improvement in the signal-to-noise ratio of the license plate's valid area, and the compensation coefficient (e.g., 0.25 to 0.28) and shrinkage factor (e.g., 0.6 to 0.65) are iteratively adjusted. When the contrast matrix gradient variance drops to the preset threshold of 0.05, the optimization is stopped and a stable license plate image is output.

[0142] In summary, this method adaptively adjusts the fill light timing and filtering parameters based on the real-time ambient light fluctuation characteristics, so that the system always works in the best anti-interference state. It also improves the signal-to-noise ratio of the target signal by intelligently staggering the interference beats and optimizing the passband characteristics, thereby ensuring that the imaging quality remains stable and reliable when the ambient light changes rapidly.

[0143] In this embodiment, the image forming method further includes the following steps:

[0144] S302: Each image acquisition device uses a single-photon avalanche diode to capture the arrival time interval of ambient light photons in real time, generates a quantized time jitter spectrum, and constructs a joint jitter spectrum array after aligning the time references using a clock synchronization protocol between the image acquisition devices.

[0145] S304: Performing a Weiner-Schinchin transform on the joint jitter spectrum array to extract the non-stationary cross-entropy characteristics of its phase noise, and generating an interference factor matrix representing the intensity of ambient light quantum interference between devices through complex domain convolution;

[0146] S306: When the off-diagonal norm of the interference factor matrix exceeds the quantum decoherence threshold, the optimal phase compensation angle is calculated according to the Poisson distribution fitting curve of the photon time jitter, and the programmable delay line is driven to generate an anti-decoherence coding timing orthogonal to the ambient light quantum fluctuation beat;

[0147] S308: Modulating the carrier envelope phase of the pulse light source at an optimal phase compensation angle so that the fill light signals of each image acquisition device form staggered frequency combs in the time-frequency domain, and locking the comb spacing to the fractional harmonics of the main frequency of the ambient light through an optical phase-locked loop;

[0148] S310: Extract the photon statistical variance of the transient interference area in the anti-interference feature map. When the photon statistical variance value is lower than the quantum threshold, the interference pattern spatial spectrum analysis is triggered. If the spatial spectrum entropy value does not meet the coherent stability criterion, it is fed back to step S304 to iteratively optimize the phase compensation angle.

[0149] For example, in a multi-camera collaborative monitoring scenario for a highway tunnel, three image acquisition devices use single-photon avalanche diodes (SPADs) to capture the inter-arrival time intervals of ambient light photons in real time. PTP clock synchronization is used to construct a joint temporal jitter spectrum. The phase noise cross-entropy characteristics are extracted using the Wiener-Schinchin transform, generating an interference factor matrix (with an off-diagonal norm of 0.75). When the quantum decoherence threshold (0.7) is exceeded, the optimal phase compensation angle of 38° is calculated based on a Poisson distribution fit to the photon jitter curve. This is used to drive a programmable delay line to generate an orthogonal coding sequence, which modulates the carrier phase of a 940nm pulsed light source, forming a frequency comb (spacing 25 GHz) locked to the quarter harmonic of the fluorescent lamp's main frequency (100 Hz). Finally, the photon statistical variance is used to trigger iterative optimization of the spatial spectral entropy value, achieving multi-device collaborative imaging that is resistant to quantum interference.

[0150] In summary, in order to solve the problem of signal decoherence and synchronization failure caused by photon fluctuations in multiple image acquisition devices under quantum-level ambient light interference, this embodiment uses quantized time jitter analysis and phase cooperative modulation to accurately characterize and compensate for quantum-level ambient light interference between multiple devices. Time-frequency domain coding is used to form a decoherence-resistant signal structure, thereby improving the system stability and signal fidelity of multi-device collaborative imaging, ensuring high-precision synchronous acquisition and interference suppression capabilities even in extreme light quantum noise environments.

[0151] In this embodiment, the image forming method further includes the following steps:

[0152] Based on the marking of transient interference areas, the Stokes parameters of the mirror-reflected light are collected by a focal plane polarization sensor, and the degree of polarization state degradation is quantified to generate a polarization pollution feature vector.

[0153] Based on the coding timing and polarization pollution characteristic vector of the pulse light source, an alternating magnetic field time-varying waveform is generated that is strictly synchronized with the rising / falling edge of the fill light pulse, and its frequency components include odd harmonics of the pulse fundamental frequency;

[0154] An alternating magnetic field is applied to a magneto-optical crystal waveguide to induce dynamic Faraday rotation of the reflected light passing through the waveguide. The Mueller matrix trace of the output light field is monitored in real time, and the magnetic field amplitude is adjusted through a feedback loop to make the rotation angle orthogonal to the polarization contamination eigenvector.

[0155] The Stokes parameter is reconstructed on the compensated reflected light to obtain the polarization purity index. When the index is lower than the background polarization of the ambient light, the magneto-optical coefficient matching optimization module is triggered.

[0156] According to the deviation between the polarization purity index and a preset threshold, the Verdet constant compensation factor of the magneto-optical crystal is dynamically corrected until the polarization pollution eigenvector of the mirror reflection area converges to the incoherent noise level.

[0157] For example, in rainy highway monitoring scenarios, to address the strong mirror reflection interference caused by road water: a split-focus plane polarization sensor is used to collect the Stokes parameters of the license plate mirror reflection light, quantify the degree of polarization degradation (such as the polarization degree dropped to 0.3), and generate a polarization pollution feature vector. Then, based on the encoding timing of the 940nm pulse light source (rising / falling edge 1μs), an alternating magnetic field time-varying waveform (amplitude ±50mT) containing odd harmonics (such as 15kHz / 25kHz) is generated. A magnetic field is applied to the TGG magneto-optical crystal waveguide, and the output light Mueller matrix trace is monitored in real time (such as when the trace fluctuation is greater than 0.5). The magnetic field amplitude is feedback-adjusted to make the Faraday rotation angle (target rotation angle 42°) orthogonal to the pollution vector, and the compensated Stokes parameter (polarization purity index ≥ 0.85) is reconstructed. When the index is lower than the ambient background polarization degree (0.82), the Verdet constant compensation factor correction (step size 0.05) is triggered, and the compensation factor is iteratively adjusted (such as from the initial value 0.8 to the optimized value 0.93) until the polarization pollution vector in the mirror reflection area converges (fluctuation < 0.1), thereby eliminating rain and fog interference and outputting a glare-free license plate image.

[0158] It should be noted that this solution addresses the imaging distortion caused by polarization degradation in strong specular reflection environments. Specifically, through dynamic magneto-optical modulation and polarization feedback control, it effectively suppresses the interference of ambient light on the target polarization characteristics, compensates for polarization degradation caused by specular reflection in real time, improves the purity and accuracy of polarization imaging, and ensures that high-quality target polarization characteristic information can be obtained even under complex reflection conditions.

[0159] The second aspect of the present invention discloses an image acquisition device, which is applied to any of the above-mentioned image imaging methods based on the anti-ambient light interference technology, such as Figure 3As shown, the image acquisition device includes an optical component module 1, a signal processing module 2, a dynamic control module 3 and a mechanical structure module 4;

[0160] The optical component module includes:

[0161] A pulse light source unit, used for emitting a fill light signal with a specific time code, the wavelength range of which matches the tunable optical filter;

[0162] Tunable optical filter unit, used to dynamically adjust the bandpass parameters according to the ambient light intensity to suppress interference frequency bands;

[0163] A photoelectric sensor unit, configured to receive a mixed light signal reflected by a target and convert it into an electrical signal;

[0164] Ambient light sensor unit, used to monitor ambient light intensity and spectrum characteristics in real time;

[0165] like Figure 4 As shown, the mechanical structure module includes:

[0166] The protective housing 1011 is made of light-shielding material and has a heat dissipation structure, and integrates optical components and circuits;

[0167] The pan / tilt mechanism 1022 is used to adjust the pitch and horizontal angle of the device and includes an anti-shake motor and an encoder;

[0168] Interface panel 1033, providing power input, data communication and external trigger interface;

[0169] Mounting base 1044, equipped with anti-vibration bracket and quick locking device, suitable for outdoor fixed or mobile deployment;

[0170] The signal processing module includes:

[0171] Timing control unit, used to synchronize the encoding timing of the pulse light source with the camera exposure time;

[0172] A frequency domain filtering unit, used for dynamically bandpass filtering the mixed signal output by the photoelectric sensor;

[0173] A time domain difference unit is used to extract the dynamic reflection component of the fill light modulation and generate an anti-interference feature map;

[0174] An image reconstruction unit, configured to reconstruct a target intrinsic image based on a differential operation between an effective reflection area and an ambient light interference area;

[0175] The dynamic control module includes:

[0176] Pulse coding optimization unit, used to adjust the duty cycle and timing of the fill light signal according to the ambient light fluctuation frequency;

[0177] A filter parameter adjustment unit, used for dynamically optimizing the passband center frequency and bandwidth of the tunable optical filter;

[0178] A feedback calibration unit is used to iteratively optimize the light source and filter parameters based on the image contrast index.

[0179] Wherein, the pan-tilt mechanism includes:

[0180] Pitch motor, rotation range ±90°;

[0181] Horizontal rotary motor, speed adjustable from 0.1 to 30° / s;

[0182] Inertial measurement unit, used to compensate for vibration caused by wind vibration in real time.

[0183] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. An image imaging method based on anti-ambient light interference technology, characterized in that: The following steps are involved: S102: transmitting a fill light signal with a specific time code to the target scene through a pulse light source built into the image acquisition device, and simultaneously loading a dynamic bandpass parameter linked to the ambient light intensity onto the tunable optical filter; S104: using a photoelectric sensor to receive a mixed light signal reflected by the target, extracting a reflected light component containing a pulse coding feature through the tunable optical filter, and generating an anti-interference feature map with a time dimension identifier; S106: Based on the time dimension identifier of the anti-interference feature map, identifying the effective reflection area in the target scene that meets the preset reflection characteristic threshold, and marking the transient interference area caused by direct ambient light; S108: performing a differential operation on the light intensity data of the effective reflection area and the spectral characteristics of the transient interference area to reconstruct an intrinsic image of the target obscured by light; S110: Dynamically adjusting the encoding timing of the pulse light source and the bandpass response curve of the tunable optical filter according to the contrast index of the target intrinsic image and the ambient light fluctuation frequency.

2. The image forming method based on the anti-ambient light interference technology according to claim 1, characterized in that: The 102 is specifically: The image acquisition device synchronously triggers the pulse light source to emit a pulse fill light signal with a preset time code. The pulse width of the pulse fill light signal matches the camera exposure time and forms a discrete fill light period within each exposure cycle. At the same time, the fill light timing mark is recorded; The ambient light intensity is collected in real time through the light sensor, and the main frequency band of the ambient light is extracted based on the ambient light spectrum characteristics to generate dynamically updated interference light bandpass suppression parameters; Calculating the degree of spectral overlap between the main frequency band of the ambient light and the center frequency band of the fill light signal, and dynamically adjusting the bandpass range of the tunable optical filter according to the spectral overlap, so that the passband center frequency of the tunable optical filter is locked to the fill light signal frequency band, and the bandwidth is adaptively contracted with the ambient light intensity to suppress interference; The fill light timing mark is used to extract the sub-frame image corresponding to the fill light period from the acquired image. The time domain difference operation is performed based on the filtered frequency domain features to eliminate the constant component of the ambient light in the time domain and retain the dynamic component of the fill light signal modulation. The spatial domain analysis of the sub-frame image after difference is performed to obtain the residual noise area. Mask matching is performed based on the spatial distribution template of the fill light signal to eliminate the low-frequency diffuse noise related to the ambient light and output the anti-interference target image.

3. The image forming method based on the anti-ambient light interference technology according to claim 1, characterized in that: The 104 is specifically: The photoelectric sensor receives the mixed light signal reflected by the target, which contains the self-pulse fill light component and the ambient light interference component; Inputting the mixed optical signal into a tunable optical filter, the tunable optical filter performs frequency domain filtering based on preloaded dynamic bandpass parameters to selectively pass a central frequency band containing pulse coding features and suppress a main frequency band of ambient light, and outputs a filtered reflected light signal; Based on the filtered reflected light signal, using the fill light timing mark recorded by the pulse light source, extracting a subframe image sequence corresponding to the fill light period; Performing a time domain difference operation on the subframe image sequence, eliminating the constant component of the ambient light in the time domain by calculating the pixel difference between adjacent subframes, and retaining the dynamic reflection component modulated by the pulse signal; Based on the dynamic reflection component, an anti-interference feature map with a time dimension identifier is generated, and the feature map represents the reflection characteristic change of the target in a time axis encoding manner.

4. The image forming method based on the anti-ambient light interference technology according to claim 3, characterized in that: Performing a time domain difference operation on the subframe image sequence, by calculating the pixel difference between adjacent subframes, eliminates the constant component of the ambient light in the time domain, and retains the dynamic reflection component modulated by the pulse signal, specifically: Based on the timing mark of the pulse light source, the sub-frame image sequence is spatially aligned to eliminate pixel offset caused by target motion or camera shake, and the brightness baseline of each sub-frame is unified to eliminate the DC bias of ambient light; According to the coding order of the fill light pulse, the difference values ​​of the corresponding pixels between adjacent subframes are calculated to generate a time domain gradient map, in which the constant component of the ambient light is offset by the difference operation, while the dynamic reflection component of the pulse modulation is retained as the gradient extreme value; The time domain gradient image is spatially filtered, and the spatial distribution template of the pulse signal is used to match the effective reflection area to suppress the random noise caused by ambient light fluctuations. At the same time, the gradient extreme value synchronized with the pulse timing is amplified through nonlinear gain. The enhanced gradient extreme value is reversely integrated according to the pulse coding timing to reconstruct the dynamic reflection intensity curve of the target during the fill light period, and the drift error accumulated during the integration process is eliminated to output a pure pulse modulated reflection signal.

5. The image forming method based on the anti-ambient light interference technology according to claim 1, characterized in that: The 106 is specifically: Based on the time dimension identification of the anti-interference feature map, the temporal change rate of its reflection intensity is calculated pixel by pixel along the time axis to generate a time domain response intensity matrix; Based on the time-domain response intensity matrix, pixels are divided into a set of candidate reflection regions with similar time-domain response patterns through cluster analysis, and a set of abnormal regions with discrete response patterns are marked; For the candidate reflection area set, calculate the average reflection response intensity of each area at the rising and falling edges of the fill light pulse; when the average reflection response intensity is greater than the preset reflection characteristic threshold, determine it as a valid reflection area and record its area boundary coordinates; Perform time domain gradient analysis on the abnormal region set to extract the peak value of its reflection intensity mutation within a single exposure cycle; If the sudden change peak has no correlation with the pulse light source coding timing, and the peak width is smaller than the preset transient interference judgment threshold, it is marked as a transient interference pulse area; The spatial proximity between the center coordinates of the transient interference pulse area and the boundary coordinates of the effective reflection area is calculated; if the distance between the two is less than the ambient light diffuse reflection radius, the transient interference pulse area is associated with the direct ambient light and marked as the direct ambient light transient interference area; The boundary coordinates of the effective reflection area and the mask coordinates of the ambient light direct transient interference area are integrated to generate a region identification map with classification labels.

6. The image forming method based on the anti-ambient light interference technology according to claim 1, characterized in that: The 108 is specifically: Based on the effective reflection area boundary coordinates and transient interference area mask coordinates of the area identification map, a pixel-level spatial mapping relationship is established to generate a dual-channel area positioning matrix with topological markings; The dynamic reflection component intensity data of the effective reflection area is used as the amplitude parameter, and the main frequency band spectrum characteristics of the ambient light in the transient interference area are used as the phase parameter to construct a complex feature vector with spatial correlation. In the complex domain, the complex eigenvectors of the effective reflection area and the complex eigenvectors of the transient interference area are conjugate-dot-multiplied. The coherence component of the ambient light interference is solved through the amplitude-phase separation mechanism to generate a decorrelation difference feature map. Performing a Hilbert transform on the decorrelation difference feature map to extract the instantaneous phase of its analytical signal, and reconstructing the continuous phase distribution of the target surface reflectivity to generate a phase gradient field; Performing a spatial convolution operation on the phase gradient field and the dynamic reflection component of the effective reflection area, suppressing the frequency domain aliasing noise in the convolution process by using Wiener filtering, and outputting the target intrinsic reflectivity basis function; According to the mask coordinates of the transient interference area, Laplace smoothing interpolation is performed on the boundary area of ​​the target intrinsic reflectivity basis function that overlaps with the ambient light diffuse reflection radius to eliminate the edge diffraction effect and finally reconstruct the target intrinsic image that is obscured by light.

7. The image forming method based on the anti-ambient light interference technology according to claim 1, characterized in that: The 110 is specifically: Calculate the local contrast distribution matrix from the target intrinsic image, and simultaneously obtain the real-time fluctuation spectrum of the ambient light sensor to extract its dominant fluctuation frequency and harmonic components; The phase compensation coefficient of the pulse light source coding timing and the filter bandpass contraction factor are generated according to the mean gradient of the local contrast distribution matrix and the dominant fluctuation frequency of the ambient light; Reconstructing the rising and falling edge timings of the pulse light source based on the phase compensation coefficient, so that the duty cycle of the fill light pulse is dynamically scaled inversely proportional to the ambient light fluctuation frequency, and inserting an anti-interference time window that is staggered with the ambient light fluctuation beat in each exposure cycle; The passband slope of the tunable optical filter is adjusted by using the filter bandpass contraction factor, so that the center frequency of the passband slope of the tunable optical filter remains consistent with the spectrum of the fill light signal, while the passband width is adaptively compressed according to the intensity of the harmonic component of the ambient light, and the roll-off distortion at the passband edge is suppressed; The adjusted pulse coding timing and filter response curve are input into the time domain verification module of the anti-interference characteristic map. By calculating the signal-to-noise ratio improvement in the effective reflection area, the phase compensation coefficient and the filter bandpass contraction factor are iteratively optimized until the gradient variance of the local contrast distribution matrix reaches the preset stability threshold.

8. An image acquisition device, applied to the image imaging method based on the anti-ambient light interference technology according to any one of claims 1 to 7, characterized in that: The image acquisition device includes an optical component module, a signal processing module, a dynamic control module and a mechanical structure module; The optical component module includes: A pulse light source unit, used for emitting a fill light signal with a specific time code, the wavelength range of which matches the tunable optical filter; Tunable optical filter unit, used to dynamically adjust the bandpass parameters according to the ambient light intensity to suppress interference frequency bands; A photoelectric sensor unit, configured to receive a mixed light signal reflected by a target and convert it into an electrical signal; Ambient light sensor unit, used to monitor ambient light intensity and spectrum characteristics in real time; The mechanical structure module includes: The protective shell is made of light-shielding material and has a heat dissipation structure, with integrated optical components and circuits; The pan / tilt mechanism is used to adjust the pitch and horizontal angle of the device, including an anti-shake motor and encoder; Interface panel, providing power input, data communication and external trigger interface; The mounting base is equipped with an anti-vibration bracket and a quick-locking device, suitable for outdoor fixed or mobile deployment.

9. The image acquisition device according to claim 8, characterized in that: The signal processing module includes: Timing control unit, used to synchronize the encoding timing of the pulse light source with the camera exposure time; A frequency domain filtering unit, used for dynamically bandpass filtering the mixed signal output by the photoelectric sensor; A time domain difference unit is used to extract the dynamic reflection component of the fill light modulation and generate an anti-interference feature map; An image reconstruction unit, configured to reconstruct a target intrinsic image based on a differential operation between an effective reflection area and an ambient light interference area; The dynamic control module includes: Pulse coding optimization unit, used to adjust the duty cycle and timing of the fill light signal according to the ambient light fluctuation frequency; A filter parameter adjustment unit, used for dynamically optimizing the passband center frequency and bandwidth of the tunable optical filter; A feedback calibration unit is used to iteratively optimize the light source and filter parameters based on the image contrast index.

10. The image acquisition device according to claim 8, characterized in that: The pan-tilt mechanism comprises: Pitch motor, rotation range ±90°; Horizontal rotary motor, speed adjustable from 0.1 to 30° / s; Inertial measurement unit, used to compensate for vibration caused by wind vibration in real time.

Images