CN-122024277-A - Intelligent recognition method and system for wearing states of multiple safety devices on construction site

CN122024277ACN 122024277 ACN122024277 ACN 122024277ACN-122024277-A

Abstract

The invention discloses a method and a system for intelligently identifying wearing states of multiple safety equipment on a construction site, which relate to the technical field of image detection and comprise the steps of acquiring parameters of the construction safety equipment, dividing the safety equipment based on key protection parts of human bodies and acquiring the parameters of the construction safety equipment corresponding to the key protection parts of each human body. According to the invention, the specific identification of different parts of equipment is realized by constructing the part color mapping table, a dedicated color reference basis is provided for multi-equipment classification identification, the head area is positioned by combining the human body proportion and the reference characteristic is obtained by extracting the personnel outline, the stability of core characteristic extraction is improved, the accurate quantification of the equipment color characteristic is realized by calculating the comparison value of each part and the head reference, the scene self-adaptation of the identification standard is realized by the identification distance, the accuracy of remote identification is improved, and the comprehensiveness and the accuracy of the multi-safety equipment wearing state identification are ensured by the wearing coefficient.

Inventors

LIU DONGSHUANG
SU JUN
WANG CHAO
WANG JIULIN
WU ZHITING
ZHANG YINGSHENG

Assignees

重庆沙坪坝交通实业有限公司

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. The intelligent recognition method for the wearing state of the multi-safety equipment on the construction site is characterized by comprising the following steps of: Acquiring construction safety equipment parameters, wherein the construction safety equipment parameters comprise safety equipment color information and safety equipment wearing position information; Dividing the safety equipment based on key protection parts of a human body according to construction safety equipment parameters, and acquiring construction safety equipment parameters corresponding to each key protection part of the human body, wherein the key protection parts of the human body comprise a head part, a trunk part, a hand part and a foot part; constructing a part color mapping table based on construction safety equipment parameters corresponding to each human body key protection part, wherein the part color mapping table comprises construction safety equipment RGB color gamut color intervals corresponding to each human body key protection part; acquiring real-time image data of a construction site, wherein the image data comprises a working area and a personnel working posture; acquiring reference characteristic information according to the real-time image data of the construction site; acquiring color contrast corresponding to each human body key protection part according to the reference characteristic information and real-time image data of a construction site; acquiring a color contrast threshold corresponding to each human critical protection part based on the safety equipment identification requirement according to the part color mapping table; judging whether personnel wear safety equipment according to the color contrast and the color contrast threshold value corresponding to each critical protection part of the human body, and if not, carrying out early warning on the wearing state of the personnel safety equipment.
2. The method for intelligently identifying the wearing states of multiple safety devices on a construction site according to claim 1, wherein the step of acquiring the reference characteristic information according to the real-time image data on the construction site specifically comprises the following steps: Denoising processing is carried out on the basis of a median filtering algorithm according to the real-time image data of the construction site, and a real-time corrected image is obtained; based on human body contour proportion analysis, determining a head area proportion coefficient, wherein the head area proportion coefficient represents the proportion coefficient of the head area to the total height of the human body and is positioned at the top of the personnel area; Based on an edge detection algorithm, performing image detection on the real-time corrected image to acquire personnel profile information; intercepting the contour of the person according to the proportion coefficient of the head area to obtain the head area; acquiring H, S channel mean values and standard deviations based on HSV color space according to the real-time corrected images corresponding to the head areas; Taking H, S channel mean values as a first reference vector and H, S channel standard deviations as a second reference vector; according to the HSV color space color uniformity analysis, an H channel standard deviation threshold value and an S channel standard deviation threshold value are obtained; determining a color gamut color interval corresponding to the head region according to the position color mapping table; Judging whether safety equipment exists in the head region according to the first reference vector, the color gamut color interval corresponding to the head region, the second reference vector, the H-channel standard deviation threshold and the S-channel standard deviation threshold, and if so, acquiring an R-channel mean value, a G-channel pixel mean value and a B-channel pixel mean value of the real-time corrected image corresponding to the head region; If the first reference vector does not exceed the color gamut color interval corresponding to the head region and the second reference vector does not exceed the H channel standard deviation threshold and the S channel standard deviation threshold, the head region is worn with safety equipment; And acquiring reference characteristic information according to the R channel mean value, the G channel pixel mean value and the B channel pixel mean value of the real-time corrected image corresponding to the head region.
3. The method for intelligently identifying the wearing states of multiple safety equipment on a construction site according to claim 2, wherein the method for acquiring the color contrast corresponding to each human critical protection part according to the reference characteristic information and the real-time image data on the construction site specifically comprises the following steps: according to the human body contour proportion, positioning other key protection parts according to the human body proportion by taking the head area as a reference, and obtaining a key protection part area; Removing abnormal brightness pixels in each key protection part area based on the HSV color space according to the key protection part areas, and calculating R channel mean value, G channel pixel mean value and B channel pixel mean value of the rest pixels as color basic characteristics of the key protection part areas; based on an RGB space color contrast calculation formula, acquiring the color contrast of each key protection part area and the head area according to the color basic characteristic and the reference characteristic information of the key protection part area; Taking the color contrast of each key protection part area and the head area as the color contrast corresponding to the key protection part of the human body; Wherein, the color contrast is specifically: In the formula, Represent the first The color contrast of the individual critical guard site areas, Represents the RGB channel mean value corresponding to the head region, Represent the first RGB channel mean values of the key guard site areas.
4. The intelligent recognition method for the wearing states of multiple safety equipment on a construction site according to claim 3, wherein the method for acquiring the color contrast threshold corresponding to each human critical protection part based on the safety equipment recognition requirement according to the part color mapping table specifically comprises the following steps: acquiring a head basic color according to the reference characteristic information and the part color mapping table, wherein the head basic color represents the color type closest to the reference characteristic color space in the head safety equipment color of the part color mapping table; Based on the basic color of the head, acquiring the basic color contrast corresponding to each human body key protection part based on a part color mapping table, wherein the basic color contrast represents the contrast between the safety equipment color of each human body key protection part in the part color mapping table and the basic color of the head; acquiring image acquisition equipment parameter data, wherein the image acquisition equipment parameter data comprises image acquisition focal length and pixel size data; Acquiring the pixel height of the head region according to the real-time corrected image corresponding to the head region; acquiring an identification distance according to the image acquisition equipment parameter data and the pixel height of the head area; acquiring a color distance deviation function based on equipment test according to the image acquisition equipment parameter data; acquiring a color contrast threshold corresponding to each human body key protection part according to the basic color contrast, the color distance deviation function and the identification distance corresponding to each human body key protection part; Wherein, the recognition distance specifically is: In the formula, In order to identify the distance from the object, As the focal length of the lens is, Represents the average height of the head of an adult, In order to be of the size of the pixel, Is the pixel height of the head region.
5. The intelligent recognition method for the wearing state of multiple safety equipment on a construction site according to claim 4, wherein the acquiring the color distance deviation function based on the equipment test according to the image acquisition equipment parameter data specifically comprises: based on equipment test requirements, identifying distances of 5m, 10m, 15m, 20m and 25m as comparison test points; based on construction safety equipment parameters, selecting standard color cards of the safety equipment colors, respectively placing the standard color cards at comparison test points, collecting images and calculating actual contrast to obtain a sample data set; fitting the data based on a linear regression equation according to the sample data set to obtain a color distance deviation function; Wherein, the color distance deviation function specifically comprises: In the formula, Indicating a distance of The color contrast in the case of a color contrast, As the distance attenuation coefficient, The identification distance is represented by a number of points, Is the theoretical contrast.
6. The intelligent recognition method for the wearing state of the multiple safety devices on the construction site according to claim 5, wherein the judging whether the personnel wear the safety devices according to the color contrast and the color contrast threshold corresponding to each critical protection part of the human body specifically comprises: acquiring candidate equipment region information according to the color contrast and the color contrast threshold value corresponding to each human critical protection part, wherein the candidate equipment region represents a region with the color contrast not exceeding the color contrast threshold value; taking the ratio of the area of the candidate equipment area corresponding to each human critical protection part to the total area of the human critical protection parts as an equipment wearing coefficient; Acquiring the reference protection area of the safety equipment corresponding to each critical protection part of the human body according to construction safety equipment parameters, wherein the reference protection area of the safety equipment represents the minimum protection area of the safety equipment corresponding to each critical protection part of the human body when the safety equipment is worn; Taking the ratio of the reference protection area of the safety equipment corresponding to each human critical protection part to the standard area of the human critical protection part as a protection area identification threshold value of the human critical protection part; Judging whether the personnel at the key protection part of the human body wear safety equipment according to the equipment wear coefficient and the protection area identification threshold value corresponding to each key protection part of the human body, and if the equipment wear coefficient exceeds the protection area identification threshold value, the personnel at the key protection part of the human body wear the safety equipment; if the equipment wearing coefficient does not exceed the protection area identification threshold, personnel at the key protection part of the human body do not wear safety equipment, and early warning is carried out on the wearing state of the personnel safety equipment.
7. A construction site multi-safety equipment wearing state intelligent recognition system for realizing the recognition method as claimed in any one of claims 1-6, comprising: the main control module is used for judging whether safety equipment exists in the head area, judging whether personnel wear the safety equipment or not, acquiring reference characteristic information, acquiring color contrast corresponding to each human body key protection part according to the reference characteristic information and real-time image data of a construction site, and acquiring a color contrast threshold corresponding to each human body key protection part according to a part color mapping table and based on the safety equipment identification requirement; The information acquisition module is used for acquiring construction safety equipment parameters, dividing the safety equipment based on key protection parts of human bodies, acquiring construction safety equipment parameters corresponding to each key protection part of human bodies, constructing a part color mapping table based on the construction safety equipment parameters corresponding to each key protection part of human bodies, and acquiring real-time image data of construction sites; The reference feature recognition module is used for acquiring a real-time correction image according to real-time image data of a construction site, carrying out image detection on the real-time correction image based on an edge detection algorithm, acquiring personnel profile information, intercepting personnel profiles according to a head region proportionality coefficient, acquiring a head region, acquiring H, S channel mean values and standard deviations based on an HSV color space and according to the real-time correction image corresponding to the head region; the display module is interacted with the main control module and is used for outputting and displaying construction safety equipment parameters, construction site real-time image data, personnel outline information, reference characteristic information and color contrast corresponding to each human body key protection part.
8. The intelligent recognition system for the wearing state of multiple safety equipment on construction sites according to claim 7, wherein the main control module specifically comprises: The control unit is used for acquiring reference characteristic information according to an R channel mean value, a G channel pixel mean value and a B channel pixel mean value of the real-time corrected image corresponding to the head region, acquiring color contrast corresponding to each human critical protection part according to the reference characteristic information and real-time image data of a construction site, and acquiring a color contrast threshold corresponding to each human critical protection part based on safety equipment identification requirements according to a part color mapping table; The information receiving unit is interacted with the information acquisition module and the reference characteristic identification module and is used for receiving data and transmitting the data to the judging unit; The judging unit is used for judging whether the head area is provided with safety equipment according to the first reference vector, the color gamut color interval corresponding to the head area, the second reference vector, the H channel standard deviation threshold value and the S channel standard deviation threshold value, and judging whether personnel wear the safety equipment according to the color contrast and the color contrast threshold value corresponding to each human critical protection part.
9. The intelligent recognition system for the wearing state of multiple safety equipment on construction sites according to claim 7, wherein the information acquisition module specifically comprises: The first acquisition unit is used for acquiring construction safety equipment parameters, dividing the safety equipment based on key protection parts of the human body and acquiring construction safety equipment parameters corresponding to each key protection part of the human body; the second acquisition unit is used for constructing a position color mapping table based on construction safety equipment parameters corresponding to each human body key protection position and acquiring real-time image data of a construction site.
10. The intelligent recognition system for the wearing state of multiple safety equipment on construction sites according to claim 7, wherein the reference feature recognition module specifically comprises: The edge detection unit is used for acquiring a real-time correction image according to the real-time image data of the construction site, and performing image detection on the real-time correction image based on an edge detection algorithm to acquire personnel profile information; The reference feature recognition unit is used for intercepting the outline of the person according to the proportion coefficient of the head area, acquiring the head area, and acquiring H, S channel mean values and standard deviations based on the HSV color space and according to the real-time corrected image corresponding to the head area.

Description

Intelligent recognition method and system for wearing states of multiple safety devices on construction site Technical Field The invention relates to the technical field of image detection, in particular to an intelligent recognition method and system for wearing states of multiple safety equipment on a construction site. Background The construction site has complex environment and dense risk factors, and frequent safety accidents such as high falling, object striking, mechanical injury and the like are frequent, and the standard wearing of safety equipment (such as safety helmets, safety clothing, protective gloves, safety shoes and the like) is the last line of defense for guaranteeing the life safety of operators. The traditional construction site safety equipment wearing inspection relies on manual inspection, has the problems of low efficiency, limited coverage range, easy influence by human negligence and the like, and is difficult to realize all-weather and omnibearing real-time supervision. Along with the transition of the construction industry to intellectualization and digitalization, intelligent identification of the wearing state of the safety equipment becomes a key means for improving the safety management level of the construction site. The accurate and efficient intelligent recognition technology not only can discover the behavior of not wearing the safety equipment normally in real time and early warn in time, reduce the accident occurrence probability, but also can lighten the workload of management staff, and realize the digital tracing and closed-loop management and control of the safety management. At present, the wearing state of multiple safety equipment on a construction site is identified, whether the real-time image cannot be accurately identified, whether the safety equipment is worn or not cannot be quickly identified according to the real-time image, the safety equipment (such as a safety helmet and a safety suit) at different positions is always identified by adopting unified identification logic, so that the identification accuracy is insufficient, and when the safety equipment is identified, if the safety equipment is identified by colors, a fixed color threshold value is always adopted, misjudgment is easy to occur, but if the safety equipment is identified by accurately identifying the image, the problems that the detection target distance is relatively short, shielding is avoided, the identification efficiency is relatively low, and the wearing state of the safety equipment cannot be quickly and accurately identified exist. Disclosure of Invention In order to solve the technical problems, the technical scheme solves the problems that whether real-time images cannot be accurately identified, whether safety equipment is worn or not cannot be quickly identified according to the real-time images, unified identification logic is adopted for safety equipment (such as safety helmets and safety wear) at different positions in the prior art, so that the identification accuracy is insufficient, and when the safety equipment is identified by colors, a fixed color threshold is adopted, misjudgment is easy to occur, but when the safety equipment is identified by the accurate identification of the images, the detection target distance is relatively short, shielding is avoided, the identification efficiency is relatively low, and the wearing state of the safety equipment cannot be quickly and accurately identified. In order to achieve the above purpose, the invention adopts the following technical scheme: a method for intelligently identifying wearing states of multiple safety equipment on a construction site comprises the following steps: Acquiring construction safety equipment parameters, wherein the construction safety equipment parameters comprise safety equipment color information and safety equipment wearing position information; Dividing the safety equipment based on key protection parts of a human body according to construction safety equipment parameters, and acquiring construction safety equipment parameters corresponding to each key protection part of the human body, wherein the key protection parts of the human body comprise a head part, a trunk part, a hand part and a foot part; constructing a part color mapping table based on construction safety equipment parameters corresponding to each human body key protection part, wherein the part color mapping table comprises construction safety equipment RGB color gamut color intervals corresponding to each human body key protection part; acquiring real-time image data of a construction site, wherein the image data comprises a working area and a personnel working posture; acquiring reference characteristic information according to the real-time image data of the construction site; acquiring color contrast corresponding to each human body key protection part according to the reference characteristic information and real-time image dat