Real-time video analysis and alarm method and system based on intelligent pod

Abstract

The present application provides a real-time video analysis and alarm method and system based on a smart pod. Among them, the method proposes a real-time video analysis and alarm method based on a smart pod. The smart pod projects a specific coded light pattern to the monitoring area, and collects the coded light signal reflected by the object to generate an image. The pattern deformation is analyzed by the optical decoding algorithm, the interference of suspended particles such as dust and fog is automatically removed, and the reflection area on the object surface is extracted. The reflection signal intensity is optimized according to the ambient light to enhance the detection accuracy. When the outline of the object exceeds the preset safety range, the alarm is immediately triggered and the monitoring terminal is notified. This method can effectively reduce environmental interference and improve monitoring accuracy and response speed. Real-time video analysis and alarm based on smart pods. The present application uses active coded light projection and intelligent decoding technology to effectively filter out environmental interference and strengthen target signals, thereby achieving accurate detection and real-time alarm of abnormal behavior of monitored objects.

Description

Technical Field

[0001] The present application relates to the field of real-time video analysis and alarm technology, and in particular to a real-time video analysis and alarm method and system based on an intelligent pod. Background Art

[0002] In adverse weather conditions such as rain, fog, and dust, traditional video surveillance systems often suffer from image blur, target loss, and high false alarm rates due to scattering and obstruction by suspended particles such as raindrops, fog droplets, and dust. For these scenarios, there is an urgent need for intelligent, real-time video analysis technology that can effectively filter out environmental interference, accurately identify surveillance targets such as vehicles, pedestrians, or unusual behavior, and quickly trigger alarms when targets cross boundaries or exhibit unusual appearances, ensuring the reliability and real-time performance of the surveillance system.

[0003] Currently, a typical solution is based on infrared imaging and dynamic background modeling. This solution uses infrared cameras to reduce the scattering effects of rain, fog, and dust on visible light, combines it with a dynamic background difference algorithm to extract moving targets, and uses morphological filtering to suppress noise interference, ultimately achieving target detection and alarm.

[0004] Although this solution can alleviate environmental interference to a certain extent, it still has obvious shortcomings: in conditions of thick fog or high-density dust, infrared penetration is limited and the target outline may still be blurred; the continuous movement of suspended particles, such as drifting dust and rain lines, can easily be misjudged as foreground targets, resulting in an increased false alarm rate; it passively relies on ambient light or infrared imaging and cannot actively enhance target signals, making it difficult to maintain stable performance in extreme weather. Summary of the Invention

[0005] The present application provides a real-time video analysis and alarm method and system based on an intelligent pod, which is used to solve the problem of poor real-time video analysis and alarm effects in the prior art.

[0006] In a first aspect, the present application provides a real-time video analysis and alarm method based on an intelligent pod, comprising:

[0007] The smart pod actively projects a predefined spatially coded light pattern into the monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image.

[0008] identifying deformation features of the spatially coded light pattern in the original coded light image using an optical decoding algorithm, and removing suspended particle interference regions in the original coded light image based on the deformation features using the optical decoding algorithm to obtain a surface reflection region of the monitored object;

[0009] Strengthening the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment;

[0010] When it is detected that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, a corresponding alarm instruction is generated and transmitted to the monitoring terminal.

[0011] Optionally, when it is detected that the contour of the enhanced reflective area on the surface of the monitored object exceeds a preset safety boundary, generating a corresponding alarm instruction and transmitting the alarm instruction to the monitoring terminal includes:

[0012] Extracting continuous edge lines composed of pixel points from the image of the reflective area on the surface of the monitored object after brightness enhancement, and closing the continuous edge lines to form an object contour polygon;

[0013] Comparing the coordinates of each vertex of the object's outline polygon with the vertex coordinates of the preset safety boundary, and generating an alarm instruction according to the comparison result;

[0014] The alarm instruction is transmitted to the monitoring terminal.

[0015] Optionally, the identifying deformation features of the spatially coded light pattern in the original coded light image using an optical decoding algorithm, and removing suspended particle interference regions in the original coded light image using the optical decoding algorithm based on the deformation features to obtain a surface reflection region of the monitored object includes:

[0016] Locating a reflection area of ​​the spatially coded light pattern in the original coded light image, and calculating a coordinate offset of each coded pixel point in the reflection area relative to a corresponding position point of the predefined spatially coded light pattern as a deformation feature;

[0017] Generate a deformation degree distribution map of the reflection area according to the deformation feature distribution;

[0018] A dynamic threshold is set based on the deformation degree distribution map, and coded pixel points with a deformation degree greater than the dynamic threshold are determined as suspended particle interference areas, and coded pixel points with a deformation degree less than or equal to the dynamic threshold are determined as reflective areas on the surface of the monitored object;

[0019] The coded pixels of the suspended particle interference area are removed from the original coded light image, and the coded pixels of the monitored object surface reflection area are retained to form a monitored object surface reflection area image.

[0020] Optionally, the smart pod actively projects a predefined spatially coded light pattern into a monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image, including:

[0021] The light source device of the intelligent pod is used to project a set of spatially coded light patterns arranged in a preset topological structure into the monitoring area;

[0022] Synchronously capturing the optical signal of the monitoring area by the image acquisition device of the smart pod to generate an initial image including the reflection area of ​​the spatially coded light pattern;

[0023] Separating a set of pixels carrying a spatially coded light pattern from the initial image, extracting a brightness fluctuation sequence of each pixel in the initial image, matching the brightness fluctuation sequence with a light and dark variation sequence of the spatially coded light pattern, retaining successfully matched pixels and discarding unmatched pixels;

[0024] The retained pixels are reassembled according to positions in the initial image to form an original coded light image containing only the reflected signal of the spatially coded light pattern.

[0025] Optionally, enhancing the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment includes:

[0026] Monitor the ambient light intensity value in real time and determine the brightness enhancement coefficient according to the current ambient light intensity value;

[0027] For each pixel in the image of the reflective area on the surface of the monitored object, an enhanced brightness value is calculated according to the brightness enhancement coefficient, and the enhanced brightness value is used to replace the original brightness value;

[0028] Outputs an image of the reflective area on the surface of the monitored object with enhanced brightness.

[0029] Optionally, comparing the coordinates of each vertex of the object outline polygon with the vertex coordinates of the preset safety boundary, and generating an alarm instruction according to the comparison result, includes:

[0030] Reading the horizontal and vertical values ​​of each vertex of the object outline polygon in the image coordinate system, and simultaneously obtaining the horizontal and vertical value ranges of all vertices of the preset safety boundary;

[0031] The horizontal and vertical values ​​of all vertices are compared with the minimum and maximum horizontal and vertical values ​​of the preset safety boundary, and the vertices that meet the conditions are marked as out-of-bounds vertices according to the comparison results;

[0032] Count the number of all vertices marked as out-of-bounds and generate an alarm instruction based on the statistical results.

[0033] Optionally, removing the coded pixels of the suspended particle interference area from the original coded light image and retaining the coded pixels of the monitored object surface reflection area to form an image of the monitored object surface reflection area includes:

[0034] Create a state record table with the same size as the original coded light image, mark the coded pixels determined to be in the suspended particle interference area as the first state, and mark the coded pixels determined to be in the reflective area of ​​the monitored object as the second state;

[0035] Copying the original coded light image to generate an intermediate processed image, scanning each coded pixel of the intermediate processed image position by position, and setting the brightness value of the scanned position to a preset background brightness reference value when the mark of the scanned position is in a first state, and maintaining the original brightness value of the scanned position unchanged when the mark of the scanned position is in a second state;

[0036] The intermediate processed image after brightness adjustment is output as the image of the reflective area on the surface of the monitored object and is stored in association with the current time stamp.

[0037] In a second aspect, the present application provides a real-time video analysis and alarm system based on an intelligent pod, comprising:

[0038] A generation module, in which the smart pod actively projects a predefined spatially coded light pattern into a monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image;

[0039] an obtaining module, which uses an optical decoding algorithm to identify deformation features of the spatially coded light pattern in the original coded light image, and removes suspended particle interference areas in the original coded light image based on the deformation features using the optical decoding algorithm to obtain a surface reflection area of ​​the monitored object;

[0040] An enhancement module, which enhances the intensity of the reflection area on the surface of the monitored object according to the optical characteristics of the current environment;

[0041] The transmission module generates a corresponding alarm instruction when it detects that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, and transmits the alarm instruction to the monitoring terminal.

[0042] The intelligent pod of the present application actively projects a predefined spatially coded light pattern into the monitoring area, and at the same time collects the coded light signal reflected by the monitored object in the monitoring area based on the spatially coded light pattern to generate an original coded light image; uses an optical decoding algorithm to identify the deformation characteristics of the spatially coded light pattern in the original coded light image, and uses an optical decoding algorithm to remove the suspended particle interference area in the original coded light image based on the deformation characteristics to obtain the surface reflection area of ​​the monitored object; according to the optical characteristics of the current environment, the intensity of the surface reflection area of ​​the monitored object is enhanced; when it is detected that the contour morphology of the enhanced surface reflection area of ​​the monitored object exceeds the preset safety boundary, a corresponding alarm instruction is generated and the alarm instruction is transmitted to the monitoring terminal.

[0043] This application has the following beneficial effects:

[0044] By actively projecting specific coded light, it reduces ambient light interference, enhances target signal recognizability, and provides a baseline for subsequent deformation analysis. Based on the deformation characteristics of the coded pattern, it accurately distinguishes target reflections from suspended particle interference, effectively improving the image signal-to-noise ratio. It adaptively optimizes signal strength in the target area to ensure clear contour extraction even in low-visibility conditions. By dynamically comparing the target outline with the safety boundary, it achieves high-precision anomaly detection, reduces false alarm rates, and improves alarm response speed.

[0045] Furthermore, in the original coded light image, the present application generates a deformation feature distribution map by calculating the coordinate offset of the coded pixel points, and separates the suspended particle interference and the target reflection area based on a dynamic threshold; in the enhanced target image, the closed contour polygon is extracted, and the alarm is triggered by comparing the vertex coordinates with the preset safety boundary.

[0046] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0048] Figure 1 A flowchart of a real-time video analysis and alarm method based on a smart pod provided by the present application is shown;

[0049] Figure 2 The figure shows a schematic diagram of the structure of a real-time video analysis and alarm system based on an intelligent pod provided by the present application;

[0050] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0051] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0052] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0053] In the field of intelligent monitoring in harsh environments such as rain, fog, and dust, existing technical solutions based on infrared imaging and dynamic background modeling have three key flaws: First, infrared imaging has limited penetration in dense fog or high-density dust, resulting in blurred target outlines; second, dynamic background modeling can easily misinterpret the continuous movement of suspended particles as foreground targets, resulting in a high false alarm rate; and most importantly, existing technologies rely entirely on passive imaging modes and lack active anti-interference mechanisms, making it difficult to maintain stable monitoring performance in extreme weather conditions. These shortcomings fundamentally stem from the existing solutions' passive adaptation to complex optical environments and their static processing of suspended particle interference. A new monitoring method that can actively enhance target signals and intelligently distinguish interference is urgently needed.

[0054] In response to the above-mentioned technical bottlenecks, the present invention proposes a real-time video analysis method for intelligent pods based on active coded light projection. This method innovatively uses predefined spatially coded light patterns to actively illuminate the monitoring area, and establishes a dynamic anti-interference mechanism by decoding the deformation characteristics of the reflected light: first, the geometric distortion characteristics of the coded light pattern are used to accurately distinguish between target surface reflection and suspended particle scattering, thereby achieving pixel-level interference filtering; then, the target signal strength is adaptively enhanced according to the environmental optical parameters; and finally, an alarm is triggered by sub-pixel comparison of the contour polygon with the safety boundary. This solution groundbreakingly combines active optical coding technology with intelligent image analysis, which not only solves the problem of insufficient penetration of infrared imaging, but also completely avoids the risk of misjudgment of background modeling through the active feature marking of coded light. Experiments show that in a dusty environment with low visibility, this method can still maintain a high target recognition accuracy, and the false alarm rate is lower than that of traditional technologies, achieving a revolutionary performance improvement of the monitoring system in harsh environments.

[0055] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0056] Figure 1 A flowchart of a real-time video analysis and alarm method based on an intelligent pod is provided for the embodiment of the present application, as shown in FIG. Figure 1 As shown, the method includes:

[0057] 101. The smart pod actively projects a predefined spatially coded light pattern into a monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image.

[0058] Optionally, in step 101, the smart pod actively projects a predefined spatially coded light pattern into a monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern, to generate an original coded light image, which may specifically include:

[0059] 1011. Use the light source device of the intelligent pod to project a set of spatially coded light patterns arranged in a preset topological structure into the monitoring area;

[0060] 1012. Synchronously capture, by means of an image acquisition device of the intelligent pod, light signals of the monitored area to generate an initial image including a reflection area of ​​the spatially coded light pattern;

[0061] 1013. Separate a set of pixels carrying a spatially coded light pattern from the initial image, extract a brightness fluctuation sequence of each pixel in the initial image, match the brightness fluctuation sequence with a light and dark variation sequence of the spatially coded light pattern, retain successfully matched pixels, and discard unmatched pixels.

[0062] 1034. Recombine the retained pixels according to their positions in the initial image to form an original coded light image containing only the reflected signal of the spatially coded light pattern.

[0063] In the above scheme, a spatially coded light pattern (SCL) is a set of specially arranged light spots or stripes, such as a checkerboard or wavy lines, projected by the smart pod's lighting. The brightness variations in each area are designed according to predetermined rules and are used to mark the monitored area. A brightness fluctuation sequence is a record of the brightness and darkness of each small pixel in the image over time, such as "bright-dark-bright," which is used to match the changing pattern of the projected light. A raw coded light image is a pure image consisting only of pixels that successfully match the coded light pattern, free of ambient light.

[0064] In an embodiment of the present application, first, through step 1011, the intelligent pod light source device is arranged according to a preset topological structure rule, for example, an 8×6 grid matrix is ​​used to bind a unique coding sequence to each topological unit, such as the unit "E3" sequence [0, 255, 128]. The topological unit coordinates are mapped to physical reflection points through a digital micromirror array, and the micromirrors are driven to reflect light according to the coding sequence to form a spatially coded light pattern carrying a spatial position mark; the spatially coded light pattern projection process is based on a preset topological structure spacing, such as 40cm×30cm real-time optical calibration, wherein the laser ranging feedback projection distance and the zoom lens is adjusted according to a proportional coefficient, such as 1.2 times, to ensure that the physically imaged spatially coded light pattern maintains the preset topological structure spacing; the synchronous control system triggers the pod camera exposure at the coding sequence switching moment of the spatially coded light pattern, such as the brightness jump point t=10ms, to achieve dual precise coverage of the spatially coded light pattern in the time dimension and the spatial dimension.

[0065] Secondly, at the moment when the intelligent pod light source projects the spatially coded light pattern, the synchronous pulse generator sends a hardware trigger signal to the image acquisition device through step 1012, driving the camera shutter to start exposure at the phase switching critical point of the coding sequence, such as the brightness jump moment t=10ms; the optical signal of the spatially coded light pattern reflected by the surface of the object in the monitoring area includes topological units, such as the real-time light intensity value of the grid intersection "E3", which is focused to the CMOS sensor through the optical lens group, and the photoelectric conversion unit converts the light signal intensity reflected by each topological unit into a digital grayscale value according to a preset sampling frequency, such as 120fps; the sensor array outputs the original electrical signal according to the spatial distribution relationship of the topological structure, generates a two-dimensional pixel matrix after AD conversion, and finally forms an initial image containing the complete spatially coded light pattern reflection area, whose pixel coordinates are strictly mapped to the topological unit position, such as the reflection point of the "E3" unit is imaged at the image coordinate (105,72)).

[0066] Next, in step 1013, pixel blocks of all suspected topological unit reflection points are extracted using an edge sharpening algorithm based on the initial image containing the spatially coded light pattern reflection area, such as connected areas with brightness greater than 200, to generate a set of pixels carrying the spatially coded light pattern. Then, for each pixel in the set, its brightness fluctuation sequence in multiple frames of the initial image is extracted, such as the brightness values ​​[0, 255, 128, 0, ...] of the pixel coordinate (105, 72) in 10 frames. The brightness fluctuation sequence of each pixel is matched against a preset brightness variation sequence of the topological unit corresponding to the spatially coded light pattern, such as the reference sequence [0, 255, 128] of unit "E3", for frame-by-frame difference matching. If the sum of the sequence deviations is lower than a threshold, the pixel is determined to be a successfully matched pixel. Finally, all successfully matched pixels are retained to form a filtered set, and unmatched pixels are eliminated, such as the ambient reflection point sequence [255, 255, 0] where the deviation from any reference sequence is greater than 50%.

[0067] Finally, in step 1014, based on the set of retained pixels that have successfully matched, for example, the coordinate (105, 72) determined in the previous step to meet the tolerance requirement of the light-dark variation sequence [0, 255, 128], the system performs a two-dimensional matrix reorganization according to the original coordinate position of each pixel in the initial image, for example, the projection mapping point position of the topological unit corresponding to the 3rd row and 5th column of (105, 72); the grayscale value of each retained pixel adopts its latest brightness state, for example, the brightness 128 at t=20ms is filled to the corresponding coordinate of the new image, and the unassigned coordinate position, that is, the pixel area where the match failed or was discarded, is uniformly set to zero grayscale; finally, the original coded light image containing only the reflection signal of the spatially coded light pattern is generated, in which the geometric distribution of the effective pixels strictly inherits the row and column characteristics of the preset topological structure, for example, an 8×6 grid arrangement, and the brightness information of each pixel synchronously retains its phase state in the coding sequence, for example, the grayscale 128 of the point (105, 72) corresponds to the third frame of the {light-dark variation sequence}.

[0068] For example, in Area B of a logistics warehouse.

[0069] Specifically, the intelligent pod projects a 10×10 dot-matrix light pattern (with a 20cm dot pitch) onto the shelf while simultaneously capturing a surveillance image of the stacked boxes. When the dot-matrix light reflects off the surface of box C, the system extracts the brightness sequence of each pixel in each frame (for example, the pixel sequence (100, 200) is [0, 240, 130]) and compares it to a preset sequence [0, 255, 128], allowing for a ±20 deviation. After retaining matching pixels, the system outputs an image showing only the 100 light spots on the surface of box C, ignoring any interference from the shelf background.

[0070] This step actively projects recognizable light patterns and matches them with dynamic sequences to accurately separate the reflected signals from the object surface, eliminate interference from ambient light, and provide a pure coded light image for subsequent analysis, ensuring the accuracy of spatial position detection of the target object.

[0071] 102. Identify deformation features of the spatially coded light pattern in the original coded light image using an optical decoding algorithm, and remove suspended particle interference regions in the original coded light image using the optical decoding algorithm based on the deformation features to obtain a surface reflection region of the monitored object.

[0072] Optionally, in step 102, identifying deformation features of the spatially coded light pattern in the original coded light image using an optical decoding algorithm, and removing suspended particle interference regions in the original coded light image using the optical decoding algorithm based on the deformation features to obtain a reflective region on the surface of the monitored object may specifically include:

[0073] 1021. Locate a reflection area of ​​the spatially coded light pattern in the original coded light image, and calculate a coordinate offset of each coded pixel point in the reflection area relative to a corresponding position point of the predefined spatially coded light pattern as a deformation feature.

[0074] 1022. Generate a deformation degree distribution map of the reflection area according to the deformation feature distribution;

[0075] 1023. Set a dynamic threshold based on the deformation distribution map, determine coded pixels with a deformation greater than the dynamic threshold as suspended particle interference areas, and determine coded pixels with a deformation less than or equal to the dynamic threshold as reflective areas on the surface of the monitored object;

[0076] 1024. Remove the coded pixels of the suspended particle interference area from the original coded light image, retain the coded pixels of the monitored object surface reflection area, and form an image of the monitored object surface reflection area.

[0077] The process of step 1024, "removing the coded pixels of the suspended particle interference area from the original coded light image, retaining the coded pixels of the monitored object surface reflection area, and forming an image of the monitored object surface reflection area," includes: creating a state record table of the same size as the original coded light image, marking the coded pixels determined to be in the suspended particle interference area as a first state, and marking the coded pixels determined to be in the monitored object surface reflection area as a second state; copying the original coded light image to generate an intermediate processed image, scanning each coded pixel of the intermediate processed image position by position, setting the brightness value of the scanning position to a preset background brightness reference value when the scanning position is marked in the first state, and maintaining the original brightness value of the scanning position unchanged when the scanning position is marked in the second state; and outputting the intermediate processed image with the brightness value adjusted as the image of the monitored object surface reflection area, and storing it in association with the current time stamp.

[0078] In the above scheme, a spatially coded light pattern refers to an optical marker pattern with a specific geometric arrangement, such as a grid or dot matrix, projected onto the monitoring scene by a projection device. The original coded light image refers to an image of the monitoring scene containing the reflection information of the spatially coded light pattern, captured by a camera. The deformation feature refers to the set of coordinate offsets of the coded pixel points relative to the predefined pattern position, reflecting the degree of distortion of the light pattern after being disturbed by suspended particles. The deformation degree distribution map is a data map that records the deformation amount of each coded pixel point in the form of a two-dimensional matrix, with larger values ​​indicating greater distortion. The suspended particle interference area refers to the image area where the light pattern is distorted due to reflection from particles such as dust and water droplets in the air. The surface reflection area of ​​the monitored object refers to the undistorted image area formed by direct reflection of the light pattern from the object's surface. The status record table refers to a marker matrix of the same size as the original image, which is used to identify the type of each pixel point (interference / surface reflection).

[0079] In the embodiment of the present application, first, in step 1021, the system locates the reflection area of ​​all spatially coded light patterns in the original coded light image according to the preset topological structure rule, that is, the pixel block corresponding to the effective topological unit; for each coded pixel point in the reflection area, such as the actual imaging point belonging to the unit "E3", the theoretical coordinates of the corresponding topological unit in the associated predefined spatially coded light pattern are first retrieved, such as the designed position (105, 72) of "E3", and then the actual imaging position of the coded pixel point, such as the measured center of gravity coordinate (108, 70), is calculated relative to the theoretical coordinates, where the formula for calculating the coordinate offset is: ,in is the horizontal coordinate offset, is the vertical coordinate offset, is the theoretical abscissa of the topological unit in the preset spatially coded light pattern, The theoretical ordinate of the topological unit, The measured horizontal coordinate of the reflection point in the original coded light image, The measured vertical coordinate of the reflection point. Finally, the binary offset is generated As the deformation feature of the coded pixel point, its physical meaning reflects the degree of physical deformation of the spatially coded light pattern in the local area. For example, the unit "E3" area is horizontally shifted to the right by 3 pixels and vertically shifted up by 2 pixels due to airflow disturbance.

[0080] Secondly, in step 1022, the deformation characteristics of all the coded pixels in the reflection area are calculated. , first through the formula Calculate the deformation value of each point, for example, a point but ,in is the degree of deformation of a single point; create a two-dimensional grid map of the same size as the original coded light image and The value is filled in the corresponding coded pixel coordinate position, for example, the theoretical position (150,100) is stored =6.7; then perform bilinear interpolation to fill the blank area with the mean value of the neighboring points, such as around point (152,103) The mean is 4.2, so the fill value is 4.2, and the calculation formula is: ,in The interpolation point and the nearest valid point The Euclidean distance of The interpolation point and the nearest valid point Euclidean distance; finally, grayscale normalization formula Map the deformation degree to the grayscale range of 0-255, for example hour =5 is converted to grayscale value 159 to generate a deformation degree distribution map that completely covers the reflection area, where is the minimum value of the deformation degree of all points, It is the maximum value of the deformation degree of all points, where the highlighted area (grayscale ≥ 200) represents the strong deformation position, and the low-brightness area (grayscale ≤ 50) represents the stable reflection point, which intuitively presents the physical distortion gradient distribution of the spatially coded light pattern.

[0081] Then, step 1023 is performed according to the deformation degree distribution of all coded pixels in the image. Value, quantitative index of deformation degree, statistical average deformation degree of all valid points and standard deviation , through the formula Set a dynamic threshold, such as 10 points value Conclusion , the dynamic threshold is 10.44; then each coded pixel point The value is compared with the threshold in real time. > dynamic threshold is determined as the suspended particle interference area, for example =10.2>10.44 is marked as dust interference. When the dynamic threshold is less than or equal to the threshold, it is determined to be the reflective area on the surface of the monitored object, for example =3.0<10.44 is marked as a stable reflective surface, ultimately achieving accurate classification and recognition of all coded pixels.

[0082] Finally, in step 1024, a state record table with the same size as the original coded light image is created. Based on the pre-determination result, the coded pixels in the suspended particle interference area are marked as the first state, i.e., the interference mark, and the coded pixels in the monitored object surface reflection area are marked as the second state, i.e., the valid mark. The original coded light image is then copied to generate an intermediate processed image, and the state record table is scanned pixel by pixel. When a scan position is marked as the first state, the brightness value of the position is immediately set to the preset background brightness reference value of 0, i.e., pure black processing, for example, the original brightness is set from 255 to 0. When the scan position is marked as the second state, the original brightness value is maintained unchanged, for example, the brightness remains unchanged at 128. Finally, the intermediate processed image with brightness adjustment is output as the monitored object surface reflection area image and stored with an associated time stamp, for example, saved as "surface_reflect_20240730_1430.bmp", to ensure that only valid reflection signals from the object surface are retained in the image.

[0083] For example, in an industrial workshop monitoring scenario (Area A), a 10×10 dot matrix coded light pattern is projected onto the surface of the equipment (Object B). After the camera captures the original image:

[0084] Specifically, first locate the reflection lattice unit on the surface of object B and calculate its offset relative to the standard lattice. The unit coordinates are (2,3), and the calculated offset is (+1,-2). When generating the deformation distribution map, the unit (2,3) is assigned , when the distribution diagram When , the unit (5,1) value 4.1>3.2 is determined to be dust interference; then the interference unit is marked in the status table, its brightness is reset to zero in the intermediate image, and the brightness of the reflection unit on the surface of object B is retained; finally, an image containing only the clear dot matrix on the surface of object B is output and stored as "20240530_1415_reflection.jpg".

[0085] This step quantifies the light pattern deformation characteristics and dynamically distinguishes between suspended particle interference and surface reflections, effectively eliminating environmental particulate matter contamination of the monitored image and improving the integrity of surface details. The output image retains the original lighting information while removing interfering pixels, providing a clean data foundation for subsequent object recognition or measurement.

[0086] 103. Strengthen the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment;

[0087] Optionally, in step 103, enhancing the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment may specifically include:

[0088] 1031. Monitor the ambient light intensity value in real time and determine the brightness enhancement coefficient according to the current ambient light intensity value;

[0089] 1032. For each pixel in the image of the reflective area on the surface of the monitored object, calculate an enhanced brightness value according to the brightness enhancement coefficient, and replace the original brightness value with the enhanced brightness value;

[0090] 1033. Output the image of the reflective area on the surface of the monitored object after brightness enhancement.

[0091] In the above solution, the ambient light intensity value refers to the ambient light intensity data collected in real time by the photosensor. The brightness enhancement factor is the magnification factor dynamically calculated based on the ambient light intensity and used to adjust the image brightness. The original brightness value refers to the initial, unprocessed brightness value (range 0-255) of each pixel in the image of the reflective area of ​​the monitored object's surface. The enhanced brightness value is the new brightness value obtained by multiplying the original brightness value by the brightness enhancement factor.

[0092] In the embodiment of the present application, first, in step 1031, the optical sensor of the smart pod is used to collect the current ambient light intensity value in real time, and the brightness enhancement coefficient is automatically calculated by the preset piecewise function algorithm: when the ambient light intensity is ≤500 When the ambient light intensity is between 500-20000, for example, in a cloudy warehouse, a fixed enhancement factor of 1.5 is used directly; when the ambient light intensity is between 500-20000 When it is cloudy / dusk, the formula Dynamic calculation coefficients, such as measured ; When the ambient light intensity is greater than 20000 lux, such as at noon on a sunny day, is the ambient light intensity; an upper limit enhancement coefficient of 2.0 is used, and the final output is a brightness enhancement coefficient that strictly matches the current ambient light conditions.

[0093] Next, in step 1032, each pixel in the image of the reflective area on the surface of the monitored object, i.e., the original brightness value is , the system applies the formula Calculate the enhanced brightness value, where The brightness enhancement coefficient determined in the previous step ( ≥1), followed immediately by Overwrite and replace the original brightness value of the pixel, for example Time calculation , and updates the brightness of the point to 160. The whole process is executed synchronously point by point without interruption until all pixels have completed the replacement and update of the enhanced brightness value.

[0094] Finally, step 1033 is used to process the memory image data that has completed the pixel-by-pixel brightness enhancement process, that is, the brightness of all pixels. Value has overwritten the original The value is converted into a standard bitmap format, such as 24-bit BMP, through an image encoder to generate a complete image of the reflective area on the surface of the monitored object with enhanced brightness; at the same time, a time and space tag is added to generate a unique file name:

[0095] For example, in the warehouse cargo monitoring scenario (Area A):

[0096] Specifically, the light sensor first detects that the ambient light intensity is 30 , lower than L1=50 The system automatically selects coefficient K1 = 1.8. The original brightness value of the cargo label area is 85. The calculated enhancement value is 85 × 1.8 = 153. The original brightness value of the cargo metal shell is 180. The enhancement value is 180 × 1.8 = 324, which is truncated to 255. The final output image has improved clarity of the label details and the metal part is not overexposed. It is stored as "20240605_1030_enhanced.jpg".

[0097] This step effectively solves the problem of difficulty identifying surface details in dimly lit scenes through environmental adaptive brightness enhancement. While ensuring that highlight areas are not overexposed, it significantly improves the recognition of details in shadow areas, providing an optimized image foundation for subsequent object feature analysis.

[0098] 104. When it is detected that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, a corresponding alarm instruction is generated and transmitted to the monitoring terminal.

[0099] Optionally, in step 104, when it is detected that the contour of the enhanced reflective area on the surface of the monitored object exceeds a preset safety boundary, generating a corresponding alarm instruction and transmitting the alarm instruction to the monitoring terminal may specifically include:

[0100] 1041. Extracting continuous edge lines formed by pixel points from the image of the reflective area on the surface of the monitored object after brightness enhancement, and closing the continuous edge lines to form an object contour polygon;

[0101] 1042. Compare the coordinates of each vertex of the object outline polygon with the coordinates of the vertices of the preset safety boundary, and generate an alarm instruction according to the comparison result;

[0102] Among them, the process of step 1042 "comparing the coordinates of each vertex of the object contour polygon with the vertex coordinates of the preset safety boundary, and generating an alarm instruction based on the comparison result" includes: reading the horizontal value and vertical value of each vertex of the object contour polygon in the image coordinate system, and synchronously obtaining the horizontal value range and vertical value range of all vertices of the preset safety boundary; comparing the horizontal value and vertical value of all vertices with the minimum horizontal and vertical values ​​and the maximum horizontal and vertical values ​​of the preset safety boundary, and marking the qualified vertices as out-of-bounds vertices according to the comparison results; counting the number of all vertices marked as out-of-bounds, and generating an alarm instruction based on the number statistics.

[0103] 1043. Transmit the alarm instruction to the monitoring terminal.

[0104] In the above scheme, a continuous edge line refers to a boundary line formed by connecting pixels with sudden changes in brightness in the reflection image of the object's surface. An object contour polygon refers to a geometric shape formed by closed continuous edge lines, which represents the area occupied by the object in the image. A preset safety boundary refers to a pre-defined rectangular monitoring area defined by a minimum horizontal value, a maximum horizontal value, a minimum vertical value, and a maximum vertical value. An out-of-bounds vertex refers to a corner point in the object contour polygon that exceeds the safety boundary range. An alarm command refers to a digital signal containing the out-of-bounds location and time, which is used to trigger an alarm on the monitoring terminal.

[0105] In the embodiment of the present application, first, in step 1041, the brightness gradient value of each pixel in the image of the reflective area of ​​the monitored object surface after brightness enhancement is calculated:

[0106]

[0107] in, is the pixel coordinate Brightness gradient value (the higher the gradient, the more obvious the edge), For coordinates The pixel brightness value; when The threshold is used to mark edge pixels. The threshold is the adjustable edge sensitivity (default 50, range 0-255). Then, the neighboring marked points are connected to form an initial continuous edge line through the neighborhood search algorithm. For example, the coordinates (100,200)-(101,200)-(102,200) are connected to form a horizontal line segment. Finally, the discontinuity points with gaps ≤ 3 pixels are closed by linear interpolation to calculate their connection points:

[0108] Connecting Points

[0109] Force the first and last endpoints to connect to form a completely closed object outline polygon, such as a rectangular area formed by 4 vertex coordinates.

[0110] Next, step 1042 reads the horizontal value of each vertex of the object outline polygon in the image coordinate system. and vertical values Synchronously obtain the horizontal value range of all vertices of the preset safety boundary, that is, the minimum horizontal value (All safe vertices ) and the maximum horizontal value (All safe vertices ) and the vertical value range, that is, the minimum vertical value (All safe vertices ) and the maximum vertical value (All safe vertices ), perform coordinate comparison on each contour vertex to generate a vertex out-of-bounds mark:

[0111]

[0112] Count the total number of marked out-of-bounds vertices:

[0113]

[0114] According to the statistical results, when the total number of out-of-bounds vertices is greater than 1, that is, there is at least one out-of-bounds vertex, an alarm instruction is generated; when the total number of out-of-bounds vertices is 0, a safety confirmation instruction is generated.

[0115] Finally, a network connection is established with the monitoring terminal in step 1043, i.e., the TCP protocol is used to ensure reliability, and the generated alarm instruction text is converted into a binary data stream. Then, a data packet is sent through the pre-configured target IP address and port number, such as 192.168.1.100:8080. After sending, a 500-millisecond timer is started to wait for the receiver to return an acknowledgment signal. If no acknowledgment is received within the timeout, retransmission is performed, i.e., a maximum of 3 retries are performed, each with an interval of 1 second, 2 seconds, and 4 seconds. When the monitoring terminal successfully receives and parses the instruction, it is displayed in real time on the monitoring screen, for example, " A red window pops up indicating "Out of bounds (520, 300) (80, 400) (300, 650)" and the alarm is stored in the alarm log database to complete the transmission loop.

[0116] For example, in the warehouse shelf monitoring scenario (Area A):

[0117] Specifically, the outline of the box (object B) is first detected as a quadrilateral with vertex coordinates (98, 60), (105, 55), (110, 62), and (102, 65). A safety boundary is preset: X: 100-200, Y: 50-150. A comparison reveals that the vertex (98, 60) has X=98<100, marking it as out of bounds. The number of out-of-bounds points, N=1>0, is then counted, generating an alarm: , and finally the command is sent to the monitoring terminal, triggering the sound and light alarm.

[0118] This step enables automated detection of object out-of-bounds positioning, promptly identifying unusual displacements through precise spatial comparison of outlines with safe zones. Alarm instructions contain specific out-of-bounds location information, helping managers quickly identify risks and enhancing the monitoring system's proactive early warning capabilities.

[0119] The following is a complete example of steps 101 to 104:

[0120] A chemical plant deployed an intelligent pod in Area A to monitor eight Class A storage tanks (numbered T1-T8). The pod, positioned 15 meters above the tanks, projected a 6×12 checkerboard coded light pattern (50 cm spacing, a coded sequence of [0, 128, 255, 64] cycles, and a 33.3 ms period). The monitoring system had a pre-set safety boundary (image coordinate system X: 200-800, Y: 300-600). At 3:03 PM one day, ambient light suddenly dropped to 50 lux due to heavy rain. Simultaneously, tank T5 tilted and shifted due to foundation settlement, triggering the following real-time response:

[0121] First, the pod's DMD chip projects a checkerboard pattern at a 120Hz refresh rate, synchronously triggering the CMOS camera (resolution 1920×1080 at 120fps) to expose the image. The 25th frame of the original image is captured at t=15:03:00.025. The system performs sequence matching on pixel (850, 450): The brightness values ​​of this point in four consecutive frames (frames 25-28) are [5, 131, 250, 69]. The deviation from the preset sequence [0, 128, 255, 64] is ≤±5 (|131-128|=3<5). This point is considered a valid reflection point on the T5 tank and is retained. The rain reflection point (200, 500) has a sequence of [230, 235, 12, 228], which deviates by more than 200 from any other coded sequence and is therefore discarded. The resulting raw coded light image is 840 × 560 pixels, showing only 72 checkerboard reflection points on the tank surface (8 tanks × 9 points / tank).

[0122] Secondly, calculate the deformation of 9 coded points on the surface of the T5 tank: the preset theoretical point (850,450) is measured to be (853,447) due to the tilt of the tank, Δx=+3, Δy=-3, and the deformation value The δ value statistics of 72 points in the whole area show that rain disturbance caused anomalies at three points on the tank roof: point (855,310) shifted to (880,295) due to raindrop reflection (Δx=25, Δy=15, δ=29.2). The system calculated mean μ=5.7, standard deviation σ=7.1, and dynamic threshold = μ+2σ=19.9. =29.2>19.9 is marked as raindrop interference, and the brightness of the coordinate (855,310) is reset to zero; while the T5 displacement point (853,447) =4.24<19.9 The retained brightness is 131. After processing, there are 69 valid points (3 raindrop interference points are removed).

[0123] Next, the photosensor detects an ambient light level of 50 lux (less than 500 lux) and activates a fixed enhancement factor k of 1.8. The original brightness of the tank displacement point T5 (853, 447) increases from 131 to a minimum enhanced brightness of (255, 131 × 1.8) = 236. The shadow area, the tank point T2 (350, 520), changes from an original brightness of 48 to an enhanced brightness of 86. The enhancement process limits the highlight area: the stainless steel flange point (860, 455) changes from an original brightness of 249 to an enhanced value of 448, which is then truncated to 255.

[0124] Finally, extract the T5 tank outline: 9 points convex hull operation generates a 5-vertex polygon (key vertex (853,447)→(862,443)→(869,440)). Safety boundary comparison: vertex (869,440) (Out-of-bounds rate 8.6%), Y=440 is within [300,600]. After the system marks the vertex out of bounds, count An alarm command is generated and sent to the monitoring center via the industrial ring network (TCP retransmission with three retransmissions and a 500ms timeout). This triggers the following actions: a pop-up window displays the red outline of the T5 tank's out-of-bounds area on the large screen, an audible and visual alarm sounds (105dB pulse beep), the coordinates (869,440) are automatically locked, the pan / tilt zoom is activated 30x, and the alarm frame is stored in the database. The monitor then confirms the T5 tank displacement alarm within 10 seconds and immediately initiates Plan E-03: closing the inlet and outlet valves and evacuating personnel from Area B. Later inspection confirmed that foundation settlement had caused the tank to tilt by 15cm, and a leak was averted thanks to the timely warning. The system's false alarm rate: 1 false alarm occurred during 24 hours of continuous operation in heavy rain (due to a flock of birds flying overhead being misinterpreted as displacement). After optimizing the filtering algorithm, this rate was reduced to 0.05%.

[0125] Figure 2 The present application embodiment provides a structural diagram of a real-time video analysis and alarm system based on an intelligent pod, as shown in FIG. Figure 2 As shown, the system includes:

[0126] Generation module 21, the smart pod actively projects a predefined spatially coded light pattern into the monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image;

[0127] Obtaining module 22, using an optical decoding algorithm to identify deformation features of the spatially coded light pattern in the original coded light image, and removing suspended particle interference areas in the original coded light image based on the deformation features using the optical decoding algorithm to obtain a surface reflection area of ​​the monitored object;

[0128] An enhancement module 23 enhances the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment;

[0129] The transmission module 24 generates a corresponding alarm instruction when it detects that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, and transmits the alarm instruction to the monitoring terminal.

[0130] Figure 2 The real-time video analysis and alarm system based on the smart pod can perform Figure 1 The implementation principles and technical effects of the smart pod-based real-time video analysis and alarm method described in the illustrated embodiment will not be elaborated on here. The specific manner in which the various modules and units of the smart pod-based real-time video analysis and alarm system perform their operations has been described in detail in the relevant embodiments of the method and will not be elaborated on here.

[0131] In one possible design, Figure 2 The real-time video analysis and alarm system based on the smart pod of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0132] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0133] The processing component 32 is used for the above Figure 1 The embodiment provides a real-time video analysis and alarm method based on an intelligent pod.

[0134] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0135] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0136] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0137] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0138] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0139] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0140] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The embodiment shown is a real-time video analysis and alarm method based on an intelligent pod.

[0141] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0142] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0143] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A real-time video analysis and alarm method based on an intelligent pod, characterized in that: include: The smart pod actively projects a predefined spatially coded light pattern into the monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image. identifying deformation features of the spatially coded light pattern in the original coded light image using an optical decoding algorithm, and removing suspended particle interference regions in the original coded light image based on the deformation features using the optical decoding algorithm to obtain a surface reflection region of the monitored object; Strengthening the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment; When it is detected that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, a corresponding alarm instruction is generated and the alarm instruction is transmitted to the monitoring terminal; The method of using an optical decoding algorithm to identify deformation features of the spatially coded light pattern in the original coded light image, and removing suspended particle interference areas in the original coded light image using the optical decoding algorithm based on the deformation features to obtain a surface reflection area of ​​the monitored object includes: Locating a reflection area of ​​the spatially coded light pattern in the original coded light image, and calculating a coordinate offset of each coded pixel point in the reflection area relative to a corresponding position point of the predefined spatially coded light pattern as a deformation feature; Generate a deformation degree distribution map of the reflection area according to the deformation feature distribution; A dynamic threshold is set based on the deformation degree distribution map, and coded pixel points with a deformation degree greater than the dynamic threshold are determined as suspended particle interference areas, and coded pixel points with a deformation degree less than or equal to the dynamic threshold are determined as reflective areas on the surface of the monitored object; The coded pixels of the suspended particle interference area are removed from the original coded light image, and the coded pixels of the monitored object surface reflection area are retained to form a monitored object surface reflection area image.

2. The method according to claim 1, characterized in that When it is detected that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, a corresponding alarm instruction is generated and the alarm instruction is transmitted to the monitoring terminal, including: Extracting continuous edge lines composed of pixel points from the image of the reflective area on the surface of the monitored object after brightness enhancement, and closing the continuous edge lines to form an object contour polygon; Comparing the coordinates of each vertex of the object's outline polygon with the vertex coordinates of the preset safety boundary, and generating an alarm instruction according to the comparison result; The alarm instruction is transmitted to the monitoring terminal.

3. The method according to claim 1, characterized in that The smart pod actively projects a predefined spatially coded light pattern into a monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image, including: The light source device of the intelligent pod is used to project a set of spatially coded light patterns arranged in a preset topological structure into the monitoring area; Synchronously capturing the optical signal of the monitoring area by the image acquisition device of the smart pod to generate an initial image including the reflection area of ​​the spatially coded light pattern; Separating a set of pixels carrying a spatially coded light pattern from the initial image, extracting a brightness fluctuation sequence of each pixel in the initial image, matching the brightness fluctuation sequence with a light and dark variation sequence of the spatially coded light pattern, retaining successfully matched pixels and discarding unmatched pixels; The retained pixels are reassembled according to positions in the initial image to form an original coded light image containing only the reflected signal of the spatially coded light pattern.

4. The method according to claim 1, wherein The step of enhancing the intensity of the reflective area on the surface of the monitored object according to the optical characteristics of the current environment includes: Monitor the ambient light intensity value in real time and determine the brightness enhancement coefficient according to the current ambient light intensity value; For each pixel in the image of the reflective area on the surface of the monitored object, an enhanced brightness value is calculated according to the brightness enhancement coefficient, and the enhanced brightness value is used to replace the original brightness value; Outputs an image of the reflective area on the surface of the monitored object with enhanced brightness.

5. The method according to claim 2, characterized in that: Comparing the coordinates of each vertex of the object outline polygon with the vertex coordinates of the preset safety boundary, and generating an alarm instruction according to the comparison result, including: Reading the horizontal and vertical values ​​of each vertex of the object outline polygon in the image coordinate system, and simultaneously obtaining the horizontal and vertical value ranges of all vertices of the preset safety boundary; The horizontal and vertical values ​​of all vertices are compared with the minimum and maximum horizontal and vertical values ​​of the preset safety boundary, and the vertices that meet the conditions are marked as out-of-bounds vertices according to the comparison results; Count the number of all vertices marked as out-of-bounds and generate an alarm instruction based on the statistical results.

6. The method according to claim 1, characterized in that: Removing coded pixel points of the suspended particle interference area from the original coded light image, retaining coded pixel points of the monitored object surface reflection area, and forming a monitored object surface reflection area image, including: Create a state record table with the same size as the original coded light image, mark the coded pixels determined to be in the suspended particle interference area as the first state, and mark the coded pixels determined to be in the reflective area of ​​the monitored object as the second state; Copying the original coded light image to generate an intermediate processed image, scanning each coded pixel of the intermediate processed image position by position, and setting the brightness value of the scanned position to a preset background brightness reference value when the mark of the scanned position is in a first state, and maintaining the original brightness value of the scanned position unchanged when the mark of the scanned position is in a second state; The intermediate processed image after brightness adjustment is output as the image of the reflective area on the surface of the monitored object and is stored in association with the current time stamp.

7. A real-time video analysis and alarm system based on an intelligent pod, used to execute the real-time video analysis and alarm method based on an intelligent pod according to any one of claims 1 to 6, characterized in that: include: A generation module, in which the smart pod actively projects a predefined spatially coded light pattern into a monitoring area, and simultaneously collects coded light signals reflected by monitored objects in the monitoring area based on the spatially coded light pattern to generate an original coded light image; an obtaining module, which uses an optical decoding algorithm to identify deformation features of the spatially coded light pattern in the original coded light image, and removes suspended particle interference areas in the original coded light image based on the deformation features using the optical decoding algorithm to obtain a surface reflection area of ​​the monitored object; An enhancement module, which enhances the intensity of the reflection area on the surface of the monitored object according to the optical characteristics of the current environment; The transmission module generates a corresponding alarm instruction when it detects that the outline of the enhanced reflective area on the surface of the monitored object exceeds the preset safety boundary, and transmits the alarm instruction to the monitoring terminal.

8. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the real-time video analysis and alarm method based on the smart pod as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the real-time video analysis and alarm method based on the smart pod according to any one of claims 1 to 6 is implemented.

Images