Fall risk early warning method and system of speed difference automatic controller based on image dynamic analysis

Abstract

The present application relates to the technical field of image data processing, in particular to a speed difference automatic controller falling risk early warning method and system based on image dynamic analysis, comprising: acquiring a brightness image; acquiring gradient amplitude and gradient direction of each pixel point in the brightness image; determining a reflection starting point of the brightness image based on the gradient amplitude of each pixel point in the brightness image, and then acquiring a reflection starting area of the brightness image; acquiring a light intensity quantization value of adjacent pixel points of the reflection starting area, and then acquiring a safety rope path of the brightness image; acquiring a deviation center value of the safety rope path to determine whether the safety rope abnormally shakes, and acquiring a mask image of the brightness image in the case that the safety rope does not abnormally shake; acquiring a falling risk of the safety rope at the current time by using a corresponding path of the safety rope in the continuous frame mask image, and then adjusting the tension of the safety rope. The present application can improve the accuracy of judging the falling risk of the operator.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image data processing, in particular to a fall risk early warning method and system of a speed-difference self-controller based on image dynamic analysis. BACKGROUND

[0002] The speed-difference self-controller, also known as a fall arrest speed-difference device, is a high-performance fall protection equipment. When the worker moves slowly, the safety rope is slowly pulled out or retracted. When the worker falls, the safety rope is rapidly pulled out.

[0003] The existing problem: since the speed-difference self-controller is used to determine whether the worker falls by sensing the speed difference, the speed-difference self-controller may be tightened due to the worker's misoperation. Therefore, the existing technology installs a video device on the speed-difference self-controller to analyze the movement state of the safety rope during the slow descent process, so as to quickly identify the abnormal swing of the safety rope during the slow descent process, which is beneficial to the accurate judgment of the abnormal falling state of the speed-difference self-controller during the slow descent process. However, the images collected by the existing technology during the slow descent process of the safety rope are blurred, which affects the accuracy of the positioning of the safety rope and the accuracy of the judgment of the falling risk of the worker. SUMMARY

[0004] The present application provides a fall risk early warning method and system of a speed-difference self-controller based on image dynamic analysis to solve the existing problem.

[0005] The fall risk early warning method and system of a speed-difference self-controller based on image dynamic analysis provided by the present application adopts the following technical scheme:

[0006] One embodiment of the present application provides a fall risk early warning method of a speed-difference self-controller based on image dynamic analysis, which comprises the following steps:

[0007] Collecting continuous frame safety rope images, pre-processing each frame safety rope image, and obtaining the brightness image corresponding to each frame safety rope image;

[0008] Obtaining the gradient amplitude and gradient direction of each pixel point in each frame brightness image;

[0009] Based on the gradient amplitude of each pixel point in each frame brightness image, determining all reflection starting points of each frame brightness image; using the gradient amplitude and brightness value of each reflection starting point of each frame brightness image, obtaining all reflection starting regions of each frame brightness image;

[0010] Obtaining the adjacent pixel points of each reflection starting area, obtaining the light intensity quantization value of each adjacent pixel point of each reflection starting area according to the gradient direction of each adjacent pixel point of each reflection starting area and the gradient direction of each pixel point in each reflection starting area;

[0011] Obtaining all safety rope paths of each frame of brightness image based on the light intensity quantization value of each adjacent pixel point of each reflection starting area;

[0012] Obtaining the deviation center value of each safety rope path by using the light intensity quantization value of each adjacent pixel point in each safety rope path of each frame of brightness image, judging whether the safety rope abnormally swings according to the deviation center value of each safety rope path, and obtaining the mask image of each frame of brightness image in the case that the safety rope does not abnormally swing;

[0013] Obtaining the falling risk of the safety rope at the current moment by using the corresponding path of the safety rope in the continuous frame mask image;

[0014] Adjusting the tension of the safety rope in the speed difference automatic controller according to the falling risk of the safety rope at the current moment.

[0015] Further, the specific steps of the method for pre-processing each frame of safety rope image and obtaining the corresponding brightness image of each frame of safety rope image include the following steps:

[0016] Filtering each frame of safety rope image to obtain a filtered image;

[0017] Extracting the brightness value of each pixel point in the filtered image to obtain an image after extracting the brightness value;

[0018] Performing gray processing on the image after extracting the brightness value to obtain the corresponding brightness image of each frame of safety rope image.

[0019] Further, the specific steps of the method for determining all reflection starting points of each frame of brightness image based on the gradient amplitude of each pixel point in each frame of brightness image include the following steps:

[0020] Sorting the gradient amplitudes of each pixel point in each frame of brightness image in descending order to obtain a gradient amplitude sequence;

[0021] First-order difference is performed on the gradient amplitude sequence to obtain a gradient difference sequence;

[0022] Obtaining the maximum value in the gradient difference sequence, and taking the gradient amplitude corresponding to the maximum value as an endpoint, and dividing the gradient amplitude sequence into a highlight sequence and a non-highlight sequence;

[0023] Determining all pixel points corresponding to the highlight sequence as all reflection starting points of each frame of brightness image.

[0024] Further, the gradient amplitude and the brightness value of each reflection starting point of each frame of brightness image are used to obtain all reflection starting regions of each frame of brightness image, including the following specific steps:

[0025] The brightness value of each reflection starting point is updated using the gradient amplitude and the brightness value of each reflection starting point of each frame of brightness image, to obtain the updated brightness value of each reflection starting point;

[0026] Based on the updated brightness value of each reflection starting point, adjacent reflection starting points are merged to obtain all reflection starting regions of each frame of brightness image.

[0027] Further, the gradient direction of each adjacent pixel point of each reflection starting region and the gradient direction of each pixel point in each reflection starting region are used to obtain the light intensity quantization value of each adjacent pixel point of each reflection starting region, including the following specific steps:

[0028] For each adjacent pixel point of each reflection starting region, the minimum value of the difference between the gradient direction of the adjacent pixel point and the gradient direction of each pixel point in the reflection starting region is obtained, and the minimum value is determined as the minimum gradient difference value of each adjacent pixel point of each reflection starting region;

[0029] The sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting region is obtained;

[0030] The light intensity quantization value of each adjacent pixel point of each reflection starting region is obtained using the minimum gradient difference value of each adjacent pixel point of each reflection starting region and the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting region.

[0031] Further, the light intensity quantization value of each adjacent pixel point of each reflection starting region is used to obtain all safety rope paths of each frame of brightness image, including the following specific steps:

[0032] The minimum value of the light intensity quantization value of all adjacent pixel points of each reflection starting region is obtained, and the adjacent pixel point corresponding to the minimum value is taken as a starting point. The edge formed by all adjacent pixel points of each reflection starting region is traversed to obtain a light intensity quantization value sequence of the edge;

[0033] The mean value of the light intensity quantization value sequence of the edge and the standard deviation of the light intensity quantization value sequence of the edge are obtained;

[0034] Based on the mean value of the light intensity quantization value sequence of the edge and the standard deviation of the light intensity quantization value sequence of the edge, a preset edge light intensity quantization value threshold is obtained;

[0035] Quantify the sequence of edge light intensity values in the sequence node, and the element greater than the preset edge light intensity quantization threshold value as the sequence node;

[0036] The combination of adjacent sequence nodes is regarded as a node group, and the node group with the largest number of nodes is determined as the maximum node group. The adjacent pixel points corresponding to the maximum node group are incorporated into the reflection starting area, the effective extension path of each reflection starting area is obtained, and the effective extension path of each reflection starting area is determined as each safety rope path of each frame of brightness image. All safety rope paths of each frame of brightness image are obtained.

[0037] Further, the light intensity quantization value of each adjacent pixel point in each safety rope path of each frame of brightness image is used to obtain the deviation center value of each safety rope path, and whether the safety rope abnormally shakes is judged according to the deviation center value of each safety rope path. In the case where the safety rope does not abnormally shake, a mask image of each frame of brightness image is obtained, and the specific steps include the following:

[0038] For each safety rope path of each frame of brightness image, the normalized value of the sum of the light intensity quantization values of all adjacent pixel points on the path is determined as the deviation center value of each safety rope path;

[0039] Whether the sum of the deviation center values of all safety rope paths is greater than a preset deviation center value threshold is judged;

[0040] In the case where the sum of the deviation center values of all safety rope paths is greater than the preset deviation center value threshold, it is determined that the safety rope abnormally shakes, and the speed difference automatic controller is locked;

[0041] In the case where the sum of the deviation center values of all safety rope paths is less than or equal to the preset deviation center value threshold, each safety rope path is subjected to mask processing, and a mask image of each frame of brightness image is obtained.

[0042] Further, the corresponding path of the safety rope in the continuous frame mask image is used to obtain the falling risk of the safety rope at the current time, and the specific steps include the following:

[0043] Superimpose two adjacent mask images in the continuous frame mask image to obtain the corresponding path of the safety rope in the continuous frame mask image and the overlapping area of the corresponding path of the safety rope in the continuous frame mask image;

[0044] The sum of the areas of all paths corresponding to the safety rope in the continuous frame mask image and the sum of the areas of the overlapping area of the corresponding path of the safety rope in the continuous frame mask image are calculated;

[0045] obtaining a swing amplitude of the safety rope based on a sum of areas of all paths corresponding to the safety rope in the continuous frame mask image and a sum of areas of overlapping regions of the paths corresponding to the safety rope in the continuous frame mask image;

[0046] obtaining a light intensity quantization value of each path corresponding to the safety rope in the continuous frame mask image;

[0047] obtaining a mean value and a standard deviation of the light intensity quantization values of all paths corresponding to the safety rope in the continuous frame mask image based on the light intensity quantization value of each path corresponding to the safety rope in the continuous frame mask image;

[0048] obtaining a falling risk of the safety rope at the current moment by using the swing amplitude of the safety rope and the mean value and the standard deviation of the light intensity quantization values of all paths corresponding to the safety rope in the continuous frame mask image.

[0049] Further, the adjusting the tension of the safety rope in the speed-difference self-controller according to the falling risk of the safety rope at the current moment comprises the following specific steps:

[0050] collecting the tension of the safety rope at the current moment;

[0051] calculating a difference value between the tension of the safety rope at the current moment and a locking tension;

[0052] obtaining a normalized value of the falling risk of the safety rope at the current moment;

[0053] obtaining an adjustment value of the tension of the safety rope at the current moment based on the difference value between the tension of the safety rope at the current moment and the locking tension and the normalized value of the falling risk of the safety rope at the current moment;

[0054] adjusting the tension of the safety rope in the speed-difference self-controller according to the adjustment value of the tension of the safety rope at the current moment.

[0055] An embodiment of the present application provides a speed-difference self-controller falling risk early warning system based on image dynamic analysis, which comprises the following modules:

[0056] a collection and preprocessing module, which is used for collecting continuous frame safety rope images, pre-processing each safety rope image and obtaining a brightness image corresponding to each safety rope image;

[0057] The analysis module is configured to obtain gradient amplitudes and gradient directions of each pixel point in each frame of luminance image; determine all reflection starting points of each frame of luminance image based on the gradient amplitudes of each pixel point in each frame of luminance image; obtain all reflection starting regions of each frame of luminance image by using the gradient amplitudes and luminance values of each reflection starting point of each frame of luminance image; obtain adjacent pixel points of each reflection starting region, and obtain light intensity quantization values of each adjacent pixel point of each reflection starting region according to the gradient directions of the adjacent pixel points of each reflection starting region and the gradient directions of each pixel point in each reflection starting region; obtain all safety rope paths of each frame of luminance image based on the light intensity quantization values of each adjacent pixel point of each reflection starting region; obtain a deviation center value of each safety rope path by using the light intensity quantization values of each adjacent pixel point in each safety rope path of each frame of luminance image, and determine whether the safety rope abnormally swings according to the deviation center value of each safety rope path; and obtain a mask image of each frame of luminance image in the case that the safety rope does not abnormally swing; and obtain a falling risk of the safety rope at the current moment by using a corresponding path of the safety rope in the continuous mask images.

[0058] The adjusting module is configured to adjust the tensioning force of the safety rope in the speed-difference self-controller according to the falling risk of the safety rope at the current moment.

[0059] The technical scheme of the present application has the following beneficial effects: the embodiment of the present application proposes a speed-difference self-controller falling risk early warning method and system based on image dynamic analysis, which uses the different reflection characteristics of the safety rope and the background to the light to screen the real safety rope region in the image, so as to further evaluate whether the worker abnormally moves according to the movement trajectory and movement range of the safety rope, and then adjust the tensioning force of the speed-difference self-controller. The embodiment of the present application can avoid the case that the positioning accuracy of the safety rope region is low due to the motion blur caused by the movement of the safety rope, thereby improving the accuracy of the falling risk judgment of the worker. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating laborious work.

[0061] Figure 1 The step flow chart of the present application is a speed-difference self-controller falling risk early warning method based on image dynamic analysis.

[0062] Figure 2 The block diagram of the present application is a speed-difference self-controller falling risk early warning system based on image dynamic analysis. Detailed Implementation

[0063] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a fall risk warning method and system for a speed difference controller based on image dynamic analysis proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0065] The following description, in conjunction with the accompanying drawings, details the specific scheme of the image-based dynamic analysis-based fall risk warning method and system provided by the present invention.

[0066] Please see Figure 1 The diagram illustrates a flowchart of a fall risk warning method for a speed differential controller based on dynamic image analysis, according to an embodiment of the present invention. The method includes the following steps:

[0067] Step S001: Acquire consecutive frames of safety rope images, preprocess each frame of safety rope image, and obtain the brightness image corresponding to each frame of safety rope image.

[0068] It should be noted that this invention proposes a method and system for fall risk warning of a speed-differential controller based on dynamic image analysis. This method utilizes the different light reflection characteristics of the safety rope and the background to filter the actual safety rope area in the image. Furthermore, it assesses whether the worker has moved abnormally based on the safety rope's trajectory and range of motion, and then adjusts the tension of the speed-differential controller accordingly. This invention avoids the low accuracy of safety rope area positioning caused by motion blur due to safety rope movement, thereby improving the accuracy of fall risk assessment for workers.

[0069] In this embodiment, each frame of the safety rope image is filtered to obtain the filtered image; the brightness value of each pixel in the filtered image is extracted to obtain the image after extracting the brightness value; the image after extracting the brightness value is grayscaled to obtain the brightness image corresponding to each frame of the safety rope image.

[0070] Specifically: First, a high-speed camera is installed along the axis of the speed difference controller to acquire images at a frequency of 30 frames per second. The images are then transmitted to the central controller via a communication bus, thus completing the image acquisition.

[0071] Then the collected image is filtered by a Gaussian filter (Gaussian function with a standard deviation of 1.5) to suppress noise and retain edge information.

[0072] Finally, the filtered image is extracted for a brightness channel, and the brightness channel is grayed to obtain a brightness image.

[0073] Step S002: Obtain the gradient amplitude and gradient direction of each pixel point in each frame of brightness image.

[0074] Specifically, the gradient of each pixel point in the brightness image is extracted to obtain gradient data. The gradient data includes gradient amplitude and gradient direction.

[0075] Step S003: Determine all reflection starting points of each frame of brightness image based on the gradient amplitude of each pixel point in each frame of brightness image; obtain all reflection starting regions of each frame of brightness image using the gradient amplitude and brightness value of each reflection starting point of each frame of brightness image.

[0076] It should be noted that the reflection characteristics of the safety rope and the background are different, and the image is analyzed based on this to distinguish the safety rope region and the background region.

[0077] In this embodiment, the gradient amplitudes of each pixel point in each frame of brightness image are sorted in descending order to obtain a gradient amplitude sequence; the gradient amplitude sequence is first-order differentiated to obtain a gradient difference sequence; the maximum value in the gradient difference sequence is obtained, and the gradient amplitude corresponding to the maximum value is taken as an endpoint to divide the gradient amplitude sequence into a highlight sequence and a non-highlight sequence; all pixel points corresponding to the highlight sequence are determined as all reflection starting points of each frame of brightness image.

[0078] Specifically, the highlight points in the brightness image are extracted: the gradient amplitudes of all pixel points in the brightness image are sorted in descending order to obtain a descending sequence, and first-order differentiation is performed to obtain a gradient difference sequence; the maximum difference value is extracted from the gradient difference sequence, and the two gradient amplitudes corresponding to the maximum difference value are extracted from the descending sequence and taken as endpoints to split the descending sequence into two partial sequences. The sequence with larger gradient amplitude is taken as the highlight sequence, and each pixel point corresponding to the highlight sequence is taken as a reflection starting point. Thus, all reflection starting points of each frame of brightness image are obtained.

[0079] In this embodiment, the brightness value of each reflection starting point is updated using the gradient amplitude and brightness value of each reflection starting point of each frame of brightness image to obtain the updated brightness value of each reflection starting point; adjacent reflection starting points are merged based on the updated brightness value of each reflection starting point to obtain all reflection starting regions of each frame of brightness image. ​

[0080] It should be noted that, to improve the ease of locating high-brightness points, the brightness values ​​of all high-brightness points have been updated. The updated high-brightness points have significantly higher reflectivity, which enables accurate identification of the safety rope area.

[0081] Specifically, the expression for the updated brightness value of the highlighted spot is as follows:

[0082]

[0083] in, Indicates the first The brightness values ​​of the high-brightness points have been updated. Indicates the first The gradient magnitude of a high-brightness point, This represents the maximum gradient magnitude in the highlighted sequence. Indicates the first The brightness value of the high-brightness spot before the update.

[0084] It should be noted that when the updated brightness value is greater than 255, the updated brightness value of the bright spot will be recorded as 255.

[0085] After updating the brightness values ​​of all the high-brightness points, since each high-brightness point serves as a reflection starting point, the reflection starting points after updating the brightness values ​​are then merged to obtain various reflection starting regions.

[0086] Step S004: Obtain the adjacent pixels of each reflection start region. Based on the gradient direction of the adjacent pixels of each reflection start region and the gradient direction of each pixel in each reflection start region, obtain the light intensity quantization value of each adjacent pixel in each reflection start region.

[0087] It should be noted that: the outward radiation trend of the reflection initiation area is determined to obtain the reflection saliency of the pixels adjacent to the reflection initiation area. That is, the true trajectory of the safety rope and the motion blur are distinguished based on the straight extension characteristics of the safety rope and the texture direction of brightness variation. The surface of the true trajectory of the safety rope is relatively flat. When blurring occurs, the direction of reflection will change. Therefore, the true trajectory of the safety rope and the motion blur are distinguished based on the outward extension of the reflection initiation area.

[0088] In this embodiment, for each adjacent pixel point of each reflection starting region, the minimum value of the difference between the gradient direction of the adjacent pixel point and the gradient direction of each pixel point in the reflection starting region is obtained, and the minimum value of the difference is determined as the minimum gradient difference value of each adjacent pixel point of each reflection starting region; the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting region is obtained; and the light intensity quantization value of each adjacent pixel point of each reflection starting region is obtained by using the minimum gradient difference value of each adjacent pixel point of each reflection starting region and the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting region.

[0089] It should be noted that: all pixel points adjacent to the current reflection starting region are extracted, for any one of the pixel points, the minimum value of the difference between the gradient direction and the gradient direction of all pixel points in the current reflection starting region is determined, and the gradient direction flatness is determined, so as to expect that it is closer to the flatness of the safety rope. That is, the stronger the reflection effect generated by the current adjacent pixel point, and the more consistent the reflection intensity difference with the remaining non-real trajectory points in its direction, the more likely the current adjacent pixel point is the real trajectory of the safety rope.

[0090] Therefore, first, the reflection difference between the reflection starting region and the surrounding is evaluated: the lower the total gradient variation of the remaining pixel points after the current adjacent pixel point is removed from the reflection starting region, the lower the gradient deviation, and the more likely the motion trajectory is extended from the reflection starting region in the direction of the current adjacent pixel point.

[0091] Specifically, the minimum value of the difference between the gradient direction of the current adjacent pixel point and the gradient direction of each pixel point in the reflection starting region is calculated first , and then the sum of the minimum difference values of all adjacent pixel points is calculated , and finally the ratio of and is taken as the light intensity quantization value of the current adjacent pixel point .

[0092] Thus, the light intensity quantization value of each adjacent pixel point of each reflection starting region is obtained.

[0093] Step S005: based on the light intensity quantization value of each adjacent pixel point of each reflection starting region, all safety rope paths of each frame of brightness image are obtained.

[0094] It should be noted that: when the safety rope is extended, it is not always extended by a single pixel point, but the real trajectory of the safety rope is relatively consistent in direction, so it is necessary to screen the effective extension direction in the neighborhood of the reflection starting region.

[0095]

[0095] In this embodiment, the minimum value of the light intensity quantization values of all adjacent pixel points of each reflection starting region is obtained, the adjacent pixel point corresponding to the minimum value is taken as a starting point, the edge formed by all adjacent pixel points of each reflection starting region is traversed, and a sequence of light intensity quantization values of the edge is obtained; the mean value of the sequence of light intensity quantization values of the edge and the standard deviation of the sequence of light intensity quantization values of the edge are obtained; the preset edge light intensity quantization value threshold is obtained based on the mean value of the sequence of light intensity quantization values of the edge and the standard deviation of the sequence of light intensity quantization values of the edge; elements greater than the preset edge light intensity quantization value threshold in the sequence of light intensity quantization values of the edge are taken as sequence nodes; a combination of adjacent sequence nodes is taken as a node group, and a node group containing the most nodes is determined as a maximum node group; the adjacent pixel points corresponding to the maximum node group are merged into the reflection starting region, the effective extension path of each reflection starting region is obtained, the effective extension path of each reflection starting region is determined as each safety rope path of each frame of luminance image, and all safety rope paths of each frame of luminance image are obtained.

[0096] Specifically, first, the pixel points adjacent to the reflection starting region are single-pixel edges, and therefore the light intensity quantization values of the adjacent pixel points are filled into the edge formed by the adjacent pixel points. Starting from the position corresponding to the minimum light intensity quantization value on the edge, the edge is traversed in an arbitrary direction to obtain a sequence of light intensity quantization values of the edge. The mean value of the sequence of light intensity quantization values is obtained, and the standard deviation of the sequence of light intensity quantization values is calculated. Sequence nodes greater than three times the standard deviation of the mean value in the current sequence are selected, and further, the node group with the most adjacent nodes in these nodes is extracted. The node group is merged into the reflection starting region to update the reflection starting region. The updated reflection starting region is updated according to the above steps until the reflection starting region cannot be updated any more, and the effective extension path of each reflection starting region is obtained.

[0097] It should be noted that, in this embodiment, three times the standard deviation of the mean value is taken as the preset edge light intensity quantization value threshold.

[0098] For the effective extension path of any reflection starting region, it can only appear on one safety rope, and therefore the effective extension path of each reflection starting region is recorded as a safety rope path. Thus, all safety rope paths of each frame of luminance image are obtained.

[0099] Step S006: The deviation center value of each safety rope path is obtained using the light intensity quantization values of the adjacent pixel points in each safety rope path of each frame of luminance image. Whether the safety rope is abnormally swinging is determined according to the deviation center value of each safety rope path. In the case where the safety rope does not abnormally swing, a mask image of each frame of luminance image is obtained.

[0100] It should be noted that the abnormal swing of the safety rope causes the safety rope to deviate from the center position of the speed-difference governor, so firstly the deviation of the safety rope relative to the visual angle of the speed-difference governor is judged, and then the actual swing amplitude of the safety rope is obtained according to the area of the safety rope in the continuous frame images.

[0101] In the embodiment, for each safety rope path of each frame luminance image, the normalized value of the sum of the light intensity quantization values of all adjacent pixel points on the path is determined as the deviation center value of each safety rope path; it is judged whether the sum of the deviation center values of all safety rope paths is greater than a preset deviation center value threshold; in the case that the sum of the deviation center values of all safety rope paths is greater than the preset deviation center value threshold, it is determined that the safety rope has an abnormal swing, and the speed-difference governor is locked; in the case that the sum of the deviation center values of all safety rope paths is less than or equal to the preset deviation center value threshold, each safety rope path is subjected to mask processing to obtain a mask image of each frame luminance image.

[0102] It should be noted that the preset deviation center value threshold is specifically set according to specific conditions, which is not specifically limited here. In the embodiment, the preset deviation center value threshold is 0.8.

[0103] Specifically, the sum of the light intensity quantization values of all adjacent pixel points on the safety rope path is calculated first , and then the normalization function of is used to normalize , and the normalized value of is taken as the deviation center value of the safety rope path.

[0104] It is judged whether the sum of the deviation center values of all safety rope paths is greater than 0.8. If it is greater than 0.8, it is judged that the current safety rope should be tightened, so as to avoid the falling risk caused by the abnormal swing of the safety rope.

[0105] If the sum of the deviation center values of all safety rope paths is less than or equal to 0.8, the pixel points corresponding to each safety rope path in the luminance image are marked as 1, and the remaining pixel points are marked as 0 to obtain an independent mask, and then each pixel point is assigned a corresponding value, which is obtained according to the normalized value of of the path corresponding to the pixel point and the corresponding value of the pixel point in the channel (red channel). For example, if the normalized value of of the path corresponding to the pixel point is 0.8, and the corresponding value of the pixel point in the channel (red channel) is 255, the value 0.8*255 should be assigned as the false color mask result of the pixel point.

[0106] Thus, the mask image of each frame luminance image is obtained.

[0107] Step S007: Obtain the falling risk of the safety rope at the current time by using the corresponding path of the safety rope in the continuous frame mask image.

[0108] It should be noted that the swing of the safety rope is judged according to the real-time continuous frame image, and the tension of the speed difference controller is adaptively adjusted, so as to reduce the falling risk. The swing of the safety rope is judged based on the safety rope path marked in the mask image: the safety rope path marked in the current frame needs to be corresponded in the subsequent frame, so as to judge the movement trajectory of the safety rope. The safety rope swings around the fixed point. Since the speed difference controller is connected to the worker, when the worker has an abnormal change in posture (such as wind causing the worker to move), it may cause the safety rope to be released quickly, causing the worker to fall. The abnormal movement of the worker is reflected by the movement trajectory of the safety rope connected to the speed difference controller.

[0109] In this embodiment, the two adjacent mask images in the continuous frame mask image are superimposed to obtain the corresponding path of the safety rope in the continuous frame mask image and the overlapping area of the corresponding path of the safety rope in the continuous frame mask image; the sum of the areas of all the corresponding paths of the safety rope in the continuous frame mask image and the sum of the areas of the overlapping areas of the corresponding paths of the safety rope in the continuous frame image are calculated; based on the sum of the areas of all the corresponding paths of the safety rope in the continuous frame mask image and the sum of the areas of the overlapping areas of the corresponding paths of the safety rope in the continuous frame image, the swing amplitude of the safety rope is obtained; the light intensity quantization value of each path corresponding to the safety rope in the continuous frame mask image is obtained; based on the light intensity quantization value of each path corresponding to the safety rope in the continuous frame mask image, the mean and standard deviation of the light intensity quantization value of all the paths corresponding to the safety rope in the continuous frame mask image are obtained; the falling risk of the safety rope at the current time is obtained by using the swing amplitude of the safety rope and the mean and standard deviation of the light intensity quantization value of all the paths corresponding to the safety rope in the continuous frame mask image.

[0110] Specifically, first, for the two adjacent images in the continuous frame mask image, the front and rear frame images are superimposed, and the safety rope path in the previous frame image , the safety rope path with the largest overlapping area in the subsequent frame image is taken as the corresponding path of the safety rope path in the previous frame image . The subsequent frame image is iterated to judge, so as to obtain the corresponding path sequence of the safety rope in each frame image. The abnormal state change of the safety rope in the moving process is evaluated by the corresponding path sequence, and the tension of the speed difference controller is adjusted according to the risk factor, so as to realize the intelligent processing of the falling risk.

[0111] Then, the areas of all the paths in the path sequence are superimposed, and the sum of the areas of all the paths is taken and the sum of the areas of the superimposed regions is taken as the swing amplitude of the current safety rope. The smaller the sum of the areas of the superimposed regions, the greater the swing amplitude of the safety rope, and the more likely the current path sequence represents the abnormal movement range of the worker.

[0112] Further, the fluctuation of the light intensity quantization value of each safety rope path is judged. When the safety rope movement trajectory changes greatly and the movement is fast due to blurring, the light intensity quantization value of each safety rope path always maintains a high numerical level in the continuous frame images, thereby indicating that the safety rope is always in an abnormal movement state.

[0113] Specifically, the falling risk at the current time is calculated as follows:

[0114]

[0115] wherein, represents the falling risk of the safety rope at the current time, represents the sum of the areas of the superimposed regions of the corresponding paths in the continuous frame mask images, represents the sum of the areas of all the paths in the continuous frame mask images, represents the mean value of the light intensity quantization values of all the paths in the continuous frame mask images, represents the standard deviation of the light intensity quantization values of all the paths in the continuous frame mask images, represents the light intensity quantization value of each path in the continuous frame mask images. It should be noted that the light intensity quantization values of the adjacent pixel points contained in each path in the continuous frame mask images are summed up as the light intensity quantization value of each path.

[0116]

[0117] The greater this value, the more significant the falling risk up to the current time, and thus the speed difference self-controller should adjust the tightening force of the safety rope at the current time. Step S008: adjusting the tightening force of the safety rope in the speed difference self-controller according to the falling risk of the safety rope at the current time.

[0118]

[0119] ​In the embodiment, the tension of the safety rope at the current time is collected; the difference between the tension of the safety rope at the current time and the locking tension is calculated; the normalized value of the falling risk of the safety rope at the current time is obtained; the adjustment value of the tension of the safety rope at the current time is obtained based on the difference between the tension of the safety rope at the current time and the locking tension and the normalized value of the falling risk of the safety rope at the current time; and the tension of the safety rope in the speed difference self-controller is adjusted according to the adjustment value of the tension of the safety rope at the current time.

[0120] Specifically, the tension of the safety rope at the current time is collected ;

[0121] The difference between the current tension and the prior locking tension is calculated , wherein ;

[0122] The product of the difference and the normalized value of the falling risk is taken as the adjustment value of the current tension , wherein ; therefore, the tension at the current time should be adjusted to .

[0123] Please refer to Figure 2 , which shows a block diagram of a falling risk early warning system of a speed difference self-controller based on image dynamic analysis provided by an embodiment of the application, and the system comprises the following modules:

[0124] The acquisition and preprocessing module 100 is used for collecting continuous frame safety rope images, pre-processing each frame safety rope image, and obtaining the corresponding brightness image of each frame safety rope image.

[0125] The analysis module 200 is configured to obtain gradient amplitudes and gradient directions of each pixel point in each frame of luminance image; determine all reflection starting points of each frame of luminance image based on the gradient amplitudes of each pixel point in each frame of luminance image; obtain all reflection starting regions of each frame of luminance image by using the gradient amplitudes and luminance values of each reflection starting point of each frame of luminance image; obtain adjacent pixel points of each reflection starting region, and obtain light intensity quantization values of each adjacent pixel point of each reflection starting region according to the gradient directions of the adjacent pixel points of each reflection starting region and the gradient directions of each pixel point in each reflection starting region; obtain all safety rope paths of each frame of luminance image based on the light intensity quantization values of each adjacent pixel point of each reflection starting region; obtain a deviation center value of each safety rope path by using the light intensity quantization values of each adjacent pixel point in each safety rope path of each frame of luminance image, and determine whether the safety rope abnormally swings according to the deviation center value of each safety rope path; obtain a mask image of each frame of luminance image in the case that the safety rope does not abnormally swing; and obtain a falling risk of the safety rope at the current moment by using a corresponding path of the safety rope in the continuous mask images.

[0126] The adjusting module 300 is configured to adjust the tensioning force of the safety rope in the speed-difference self-controller according to the falling risk of the safety rope at the current moment.

[0127] Thus, the present application is completed.

[0128] In summary, in the embodiment of the present application, the real safety rope region in the image is screened by using the different reflection characteristics of the safety rope and the background on light, so as to further evaluate whether the abnormal movement of the worker occurs according to the movement trajectory and the movement range of the safety rope, and then adjust the tensioning force of the speed-difference self-controller. The embodiment of the present application can avoid the case that the positioning accuracy of the safety rope region is low due to the motion blur caused by the movement of the safety rope, so as to improve the accuracy of the judgment on the falling risk of the worker.

[0129] The above only describes the preferred embodiments of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A speed difference automatic controller falling risk early warning method based on image dynamic analysis, characterized in that, The method comprises the following steps: The method comprises the following steps: Collecting continuous frame safety rope images, pre-processing each frame safety rope image, and obtaining the corresponding brightness image of each frame safety rope image; Obtaining the gradient amplitude and gradient direction of each pixel point in each frame brightness image; Based on the gradient amplitude of each pixel point in each frame brightness image, determine all reflection starting points of each frame brightness image; using the gradient amplitude and brightness value of each reflection starting point of each frame brightness image, obtain all reflection starting regions of each frame brightness image; Obtaining the adjacent pixel points of each reflection starting region, and obtaining the light intensity quantization value of each adjacent pixel point of each reflection starting region according to the gradient direction of each adjacent pixel point of each reflection starting region and the gradient direction of each pixel point in each reflection starting region; Wherein, according to the gradient direction of each adjacent pixel point of each reflection starting region and the gradient direction of each pixel point in each reflection starting region, the light intensity quantization value of each adjacent pixel point of each reflection starting region is obtained, which comprises the following specific steps: For each adjacent pixel point of each reflection starting region, the minimum value of the difference between the gradient direction of the adjacent pixel point and the gradient direction of each pixel point in the reflection starting region is obtained, and the minimum value is determined as the minimum gradient difference value of each adjacent pixel point of each reflection starting region; Obtaining the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting region; The ratio of the minimum gradient difference value of each adjacent pixel point of each reflection starting region to the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting region is determined as the light intensity quantization value of each adjacent pixel point of each reflection starting region; Based on the light intensity quantization value of each adjacent pixel point of each reflection starting region, obtain all safety rope paths of each frame brightness image; Using the light intensity quantization value of each adjacent pixel point in each safety rope path of each frame brightness image, obtaining the deviation center value of each safety rope path, judging whether the safety rope is abnormally swinging according to the deviation center value of each safety rope path, and obtaining the mask image of each frame brightness image in the case that the safety rope does not abnormally swing; Using the corresponding path of the safety rope in the continuous frame mask image, obtaining the falling risk of the safety rope at the current time; 2. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 1, characterized in that, According to the falling risk of the safety rope at the current time, adjust the tension of the safety rope in the speed difference automatic controller. The specific steps of pre-processing each frame safety rope image and obtaining the corresponding brightness image of each frame safety rope image are as follows: Filtering each frame safety rope image to obtain a filtered image; Extracting the brightness value of each pixel point in the filtered image to obtain an image after extracting the brightness value; 3. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 2, characterized in that, Performing grayscale processing on the image after extracting the brightness value to obtain the corresponding brightness image of each frame safety rope image. The specific steps of determining all reflection starting points of each frame brightness image based on the gradient amplitude of each pixel point in each frame brightness image are as follows: Sort the gradient amplitudes of each pixel point in each frame brightness image in descending order to obtain a gradient amplitude sequence; First-order difference is performed on the gradient amplitude sequence to obtain a gradient difference sequence; Obtaining the maximum value in the gradient difference sequence, and taking the gradient amplitude corresponding to the maximum value as an end point, and dividing the gradient amplitude sequence into a highlight sequence and a non-highlight sequence; Determine all pixel points corresponding to the highlight sequence as all reflection starting points of each frame of luminance image.

4. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 3, characterized in that, The specific steps for obtaining all reflection starting regions of each frame of luminance image by using the gradient amplitude and luminance value of each reflection starting point of each frame of luminance image are as follows: Updating the luminance value of each reflection starting point by using the gradient amplitude and luminance value of each reflection starting point of each frame of luminance image, and obtaining the updated luminance value of each reflection starting point; Based on the updated luminance value of each reflection starting point, merging adjacent reflection starting points to obtain all reflection starting regions of each frame of luminance image.

5. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 1, characterized in that, The specific steps for obtaining all safety rope paths of each frame of luminance image based on the light intensity quantization value of each adjacent pixel point of each reflection starting region are as follows: Obtaining the minimum value of the light intensity quantization value of all adjacent pixel points of each reflection starting region, taking the adjacent pixel point corresponding to the minimum value as a starting point, and traversing the edge formed by all adjacent pixel points of each reflection starting region to obtain a light intensity quantization value sequence of the edge; Obtaining the mean value of the light intensity quantization value sequence of the edge and the standard deviation of the light intensity quantization value sequence of the edge; Based on the mean value of the light intensity quantization value sequence of the edge and the standard deviation of the light intensity quantization value sequence of the edge, obtaining a preset edge light intensity quantization value threshold; Taking elements greater than the preset edge light intensity quantization value threshold in the light intensity quantization value sequence of the edge as sequence nodes; Taking adjacent sequence nodes as a combination to form a node group, determining the node group with the largest number of nodes as the largest node group, merging the adjacent pixel points corresponding to the largest node group into the reflection starting region to obtain an effective extension path of each reflection starting region, and determining the effective extension path of each reflection starting region as each safety rope path of each frame of luminance image to obtain all safety rope paths of each frame of luminance image.

6. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 5, characterized in that, The specific steps for obtaining a mask image of each frame of luminance image in the case where the safety rope does not abnormally sway are as follows: For each safety rope path of each frame of luminance image, determine the normalized value of the sum of the light intensity quantization values of all adjacent pixel points on the path as the deviation center value of each safety rope path; Determine whether the sum of the deviation center values of all safety rope paths is greater than a preset deviation center value threshold; In the case where the sum of the deviation center values of all safety rope paths is greater than the preset deviation center value threshold, determine that the safety rope abnormally sways, and lock the speed difference controller; In the case where the sum of the deviation center values of all safety rope paths is less than or equal to the preset deviation center value threshold, perform mask processing on each safety rope path to obtain a mask image of each frame of luminance image.

7. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 6, characterized in that, The method comprises the following steps: Superimpose two adjacent mask images in the continuous frame mask images to obtain the corresponding path of the safety rope in the continuous frame mask images and the overlapping area of the corresponding path of the safety rope in the continuous frame mask images; Calculate the sum of the areas of all the corresponding paths of the safety rope in the continuous frame mask images and the sum of the areas of the overlapping areas of the corresponding paths of the safety rope in the continuous frame mask images; Obtain the swing amplitude of the safety rope based on the sum of the areas of all the corresponding paths of the safety rope in the continuous frame mask images and the sum of the areas of the overlapping areas of the corresponding paths of the safety rope in the continuous frame mask images; Obtain the light intensity quantization value of each path corresponding to the safety rope in the continuous frame mask images; Obtain the mean value and standard deviation of the light intensity quantization values of all the paths corresponding to the safety rope in the continuous frame mask images based on the light intensity quantization value of each path corresponding to the safety rope in the continuous frame mask images; Obtain the falling risk of the safety rope at the current time by using the swing amplitude of the safety rope and the mean value and standard deviation of the light intensity quantization values of all the paths corresponding to the safety rope in the continuous frame mask images.

8. The fall risk early warning method based on image dynamic analysis of the speed difference automatic controller according to claim 7, characterized in that, The method comprises the following steps: Collect the tightening force of the safety rope at the current time; Calculate the difference between the tightening force of the safety rope at the current time and the locking tightening force; Obtain the normalized value of the falling risk of the safety rope at the current time; Obtain the adjustment value of the tightening force of the safety rope at the current time based on the difference between the tightening force of the safety rope at the current time and the locking tightening force and the normalized value of the falling risk of the safety rope at the current time; Adjust the tightening force of the safety rope in the speed difference controller according to the adjustment value of the tightening force of the safety rope at the current time.

9. A speed difference automatic controller falling risk early warning system based on image dynamic analysis, characterized in that, The system comprises the following modules: The acquisition and preprocessing module is used to acquire continuous frame safety rope images, pre-process each frame of safety rope image, and obtain the corresponding brightness image of each frame of safety rope image; The analysis module is used to obtain the gradient amplitude and gradient direction of each pixel point in each frame of brightness image; Determine all the reflection starting points of each frame of brightness image based on the gradient amplitude of each pixel point in each frame of brightness image; and obtain all the reflection starting areas of each frame of brightness image by using the gradient amplitude and brightness value of each reflection starting point of each frame of brightness image. The gradient direction of each adjacent pixel point of each reflection starting area and the gradient direction of each pixel point in each reflection starting area are obtained, and the light intensity quantization value of each adjacent pixel point of each reflection starting area is obtained according to the gradient direction of each adjacent pixel point of each reflection starting area and the gradient direction of each pixel point in each reflection starting area; based on the light intensity quantization value of each adjacent pixel point of each reflection starting area, all safety rope paths of each frame of luminance image are obtained; the deviation center value of each safety rope path is obtained by using the light intensity quantization value of each adjacent pixel point in each safety rope path of each frame of luminance image, and whether the safety rope abnormally swings is judged according to the deviation center value of each safety rope path, and in the case that the safety rope does not abnormally swing, a mask image of each frame of luminance image is obtained; the falling risk of the safety rope at the current moment is obtained by using the corresponding path of the safety rope in the continuous frame mask image; wherein the specific steps of obtaining the light intensity quantization value of each adjacent pixel point of each reflection starting area according to the gradient direction of each adjacent pixel point of each reflection starting area and the gradient direction of each pixel point in each reflection starting area include the following: for each adjacent pixel point of each reflection starting area, the minimum value of the difference between the gradient direction of the adjacent pixel point and the gradient direction of each pixel point in the reflection starting area is obtained, and the minimum value of the difference is determined as the minimum gradient difference value of each adjacent pixel point of each reflection starting area; the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting area is obtained; the ratio of the minimum gradient difference value of each adjacent pixel point of each reflection starting area to the sum of the minimum gradient difference values of all adjacent pixel points of each reflection starting area is determined as the light intensity quantization value of each adjacent pixel point of each reflection starting area; The adjusting module is configured to adjust the tensioning force of the safety rope in the speed difference controller according to the falling risk of the safety rope at the current moment.

Images