A tower crane hoisted object height estimation method and device based on target recognition

Abstract

The application provides a tower crane hoisted object height estimation method and device based on target recognition, comprising: obtaining tower base hook size information and hoisted object density information; obtaining hoisted object weight, distance from hook bottom end to hoisted object top end, and hook ground clearance; obtaining hoisted object overhead video image, which contains complete hook image and hoisted object image; calculating pixel ratio of hook and hoisted object in video image; calculating hoisted object area according to hook size information, distance from hook bottom end to hoisted object top end, hook ground clearance and pixel ratio, and judging hoisted object type through target recognition, further calculating hoisted object height and hoisted object bottom ground clearance. The application realizes identification of hoisted object type, planar size and hoisted object height estimation, realizes judgment of hoisted object bottom ground clearance, provides more accurate and practical data support for tower crane safe operation, and improves safety control efficiency.

Description

Technical Field

[0001] This invention belongs to the field of intelligent tower crane technology, and relates to a method and device for estimating the height of a tower crane load based on target recognition, specifically a method and device for load recognition and height estimation based on image feature analysis. Background Technology

[0002] Tower cranes, as commonly used vertical transportation equipment on construction sites, offer advantages such as reducing labor intensity, improving production efficiency, and wide coverage. However, with the continuous expansion of tower crane operations, the risk of safety accidents remains high. The core reasons lie in management deficiencies, weak safety awareness among operators, risky operations, and unlicensed operation. Currently, tower cranes are widely used on construction sites, and safety monitoring systems and collision avoidance mechanisms are commonly installed. Installing safety monitoring systems on tower cranes provides drivers with more detailed monitoring data and visual information about the loads, reducing some potential hazards.

[0003] Current solutions can only obtain the hook height (the distance from the bottom of the hook to the ground), but cannot obtain the distance from the bottom of the suspended object to the ground.

[0004] Therefore, existing anti-collision functions can only use the bottom of the hook as the judgment boundary, and cannot achieve vertical anti-collision judgment. The actual accuracy of tower crane anti-collision is low, which may lead to safety hazards caused by excessively large loads.

[0005] Therefore, how to provide a tower crane hoist height estimation method and device based on target recognition that can determine the height of the hoisted object from the ground is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention proposes a method and apparatus for estimating the height of tower crane loads based on target recognition, thereby solving the technical problems in the prior art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] This invention discloses a method for estimating the height of a tower crane load based on target recognition, comprising the following steps:

[0009] Obtain information on the dimensions of the tower base hook and the density of the suspended load;

[0010] After obtaining the weight information of the suspended object and determining that the object has been lifted, obtain the distance from the bottom of the hook to the top of the suspended object, as well as the height of the hook above the ground;

[0011] Acquire overhead video images of the suspended object, the video images containing complete images of the hook and the suspended object; identify the boundaries of the hook and the suspended object in the video images, and calculate the pixel ratio of the hook and the suspended object in the video images;

[0012] The area of ​​the suspended object is calculated based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground, and the pixel ratio. The height of the suspended object is then calculated by combining the weight and density information of the suspended object.

[0013] Calculate the height of the bottom of the load from the ground based on the height of the load, the distance from the bottom of the hook to the top of the load, and the height of the hook from the ground.

[0014] Preferably, the dimensions of the tower base hook include hook length, hook width, and hook height.

[0015] Preferably, the steps for obtaining the density information of the suspended object include:

[0016] Pre-store different types of suspended objects and the corresponding density information for each type of suspended object;

[0017] Automatically obtain the density information of the suspended object based on the current type of suspended object.

[0018] Preferably, the step of calculating the area of ​​the suspended object based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground, and the pixel ratio includes:

[0019]

[0020] Among them, S d The top view area of ​​the hook is calculated based on the hook dimensions, where H is the distance from the boom to the ground. j H is the height of the hook. g H is the height of the hook from the ground. s S' / S is the distance from the bottom of the hook to the top of the load. d 'This refers to the pixel ratio of the hook and the suspended object in the video image.

[0021] Preferably, the steps for calculating the height of the bottom of the suspended object from the ground, based on the height of the suspended object, the distance from the bottom of the hook to the top of the suspended object, and the height of the hook from the ground, include:

[0022]

[0023] Among them, H g H is the height of the hook from the ground. s M is the distance from the bottom of the hook to the top of the load, ρ is the weight of the load, and S is the density of the load. d 'S' represents the pixel area of ​​the hook in the video image. d The top view area of ​​the hook is calculated based on the hook size information.

[0024] Preferably, the step of identifying the hook boundary and the suspended object boundary in the video image using a target recognition algorithm includes:

[0025] Each frame of video image is divided into N grid units. The boundaries of the target object to be lifted and the hook within the grid are identified, the type of object to be lifted is determined, the corresponding type density is obtained from the database, and the pixels occupied are calculated to realize the detection and positioning of the target object to be lifted and the hook.

[0026] Predict the relative coordinates of the target with respect to the bounding box of its grid cell, as well as the target label and the probability that the target appears in the cell.

[0027] This invention also discloses a tower crane lifting height estimation device based on the aforementioned target recognition-based tower crane lifting height estimation method, comprising: a controller, a camera module, and a ranging module; wherein,

[0028] The camera module is installed on the tower crane and is used to capture overhead video images of the suspended load and send them to the controller.

[0029] The ranging module is installed at different positions on the hook to obtain the distance from the bottom of the hook to the top of the suspended object, as well as the height of the hook above the ground;

[0030] The controller is used to perform the following steps: receive and store information on the size of the tower base hook and the density of the suspended object; identify the hook boundary and the suspended object boundary in the video image, and calculate the pixel ratio of the hook and the suspended object in the video image; calculate the area of ​​the suspended object based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground, and the pixel ratio, and calculate the height of the suspended object by combining the weight and density information of the suspended object; calculate the height of the bottom of the suspended object above the ground based on the height of the suspended object, the distance from the bottom of the hook to the top of the suspended object, and the hook height above the ground.

[0031] Preferably, it further includes: a display module, which is electrically connected to the controller and is used to display video images, as well as the hook boundary and the suspended object boundary in the video images.

[0032] Preferably, it further includes: a power module, which is electrically connected to the controller and is used to supply power to the controller, the camera module and the ranging module.

[0033] Preferably, the ranging module uses an infrared sensor.

[0034] As can be seen from the above technical solution, compared with the prior art, the beneficial effects of the present invention include:

[0035] This invention utilizes an image-based target recognition algorithm to determine the height of a suspended object and its bottom height from the ground using images captured by a camera, especially when it's inconvenient to measure the height of the suspended object in real time and inform the operator. This data plays a significant role in construction, greatly mitigating potential safety hazards caused by the operator's inability to monitor the object's height during transport. It also significantly enhances the accuracy and reliability of existing tower crane safety monitoring systems in identifying vertical obstacles. Because this invention replaces manual height measurement with computer-based height detection, manual measurement is unnecessary when changing to different objects, greatly saving time and money. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0037] Figure 1 A flowchart of a tower crane lifting height estimation method based on target recognition provided in one embodiment of the present invention;

[0038] Figure 2 This is a schematic diagram showing the positional relationship between the hook and the suspended object according to one embodiment of the present invention;

[0039] Figure 3 AA is a schematic diagram showing the positional relationship between the hook and the suspended object according to an embodiment of the present invention;

[0040] Figure 4 This is a schematic diagram (BB) showing the positional relationship between the hook and the suspended object according to an embodiment of the present invention;

[0041] Figure 5 A block diagram of a device for determining the height of a tower crane load based on target recognition, according to an embodiment of the present invention;

[0042] Figure 6 This is a video image acquired by a camera module according to one embodiment of the present invention;

[0043] Figure 7 This is a schematic diagram of the hook boundary and the suspended object boundary in a video image acquired by a camera module according to an embodiment of the present invention;

[0044] Figure 8 This is a schematic diagram of the installation position of a weight sensor according to one embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] The first aspect of this invention provides a method for estimating the height of a tower crane load based on target recognition, comprising the following steps:

[0047] Obtain information on the dimensions of the tower base hook and the density of the suspended load;

[0048] After obtaining the weight of the load and determining when the load is lifted, obtain the distance from the bottom of the hook to the top of the load, as well as the height of the hook off the ground;

[0049] Acquire overhead video images of the suspended object, which include complete images of the hook and the object; identify the boundaries of the hook and the object in the video images, and calculate the pixel ratio of the hook and the object in the video images;

[0050] The area of ​​the suspended object is calculated based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground, and the pixel ratio. The height of the suspended object is then calculated by combining the weight and density information of the suspended object.

[0051] Calculate the height of the bottom of the load from the ground based on the height of the load, the distance from the bottom of the hook to the top of the load, and the height of the hook from the ground.

[0052] In one embodiment, the tower base hook size information includes hook length, hook width, and hook height.

[0053] In one embodiment, the step of obtaining the density information of the suspended object includes:

[0054] Pre-store the characteristic information of different types of suspended objects, as well as the suspended object density information corresponding to each type of suspended object;

[0055] Automatically obtain the density information of the suspended object based on the current type of suspended object.

[0056] like Figure 1 As shown, in one embodiment, the planar length, width, and height of the hook are pre-configured, and the system configures the tower crane hook planar length L. g With width W g Information, configuration of hook height H j information;

[0057] Next, configure the types of objects to be hoisted in the project and their corresponding density information in the system. Then, preset the types of objects to be hoisted according to the project requirements (common construction materials such as steel bars, concrete, timber, steel pipes, etc.) and assign a density range to each preset type of object.

[0058] Next, the camera installed on the crane captures images of the suspended object, obtaining video footage, and the infrared sensor on the hook determines the hook's height H above the ground. g ;

[0059] like Figure 8 As shown, a weight sensor is installed at position A on the tower base. Connected to the pulley system, it measures the weight of the suspended load. After the hook moves upward, the sensor detects the weight of the load and determines the moment when the load is fully lifted, based on the sensor's readings. The distance H from the bottom of the hook to the top of the load is then calculated. s Simultaneously, the video transmission data is frame-by-frame extracted and the availability of image pixels is determined to identify the hook boundary. Then, the actual cross-sectional area of ​​the suspended object is calculated based on the pixel ratio of the suspended object and the hook in the image, and its thickness (i.e., the height of the suspended object) is determined based on the mass of the suspended object measured by the sensor.

[0060] It should be noted that when the boundary of the object to be hoisted exceeds the current camera's outer frame, the camera's focal length should be reduced and a new screenshot taken.

[0061] like Figure 2-4 As shown, in one embodiment, an algorithm performs target recognition during the calculation of the hanging object's pixel area, separately selecting the pixel areas of the hanging object and the hook. Since the planar area of ​​the hook is known, and considering the perspective effect caused by the height difference between the hook and the hanging object, the area of ​​the hanging object can be calculated through a certain ratio conversion. The specific calculation steps and formulas are as follows:

[0062] Given the top view area S of the hook d The distance from the boom to the ground, H; the height of the hook, H. j H, the height of the hook from the ground g The height H of the hook from the upper surface of the suspended object s Based on the pixel area S of the hook in the camera image d By calculating the ratio of the pixel area S of the suspended object to the actual area estimate S, we can obtain the actual area estimate S of the suspended object.

[0063]

[0064] We can obtain,

[0065] A frame of image is acquired from the camera; feature extraction is performed using a trained model file to obtain the target image feature information; this is compared and filtered with pre-stored suspended object feature information to identify the suspended object type, and the suspended object density ρ is obtained by comparing it with the configured suspended object density range. Treating the suspended object as a cylinder with a constant cross-sectional area, the formula for the weight of a cylinder is:

[0066] M=ρ*S*H w (2)

[0067] By combining formulas (1) and (2), the height H of the suspended object can be obtained. w

[0068]

[0069] Based on the known hook height H above the ground g and the height H of the hook from the upper surface of the load s The actual height H of the suspended object can be obtained. d :

[0070]

[0071] In one embodiment, the step of identifying the hook boundary and the suspended object boundary in a video image using a target recognition algorithm includes:

[0072] Target recognition includes identifying two very similar targets of the same type, as well as identifying one type of target with other types of targets. The specific process consists of individual or multiple tasks such as classification, localization, detection, and segmentation. This algorithm works by dividing the image into N grid cells, each with an equal-sized region. Each of these N grid cells is responsible for detecting and localizing the target within its cell. Accordingly, these grids predict X relative coordinates of the bounding box to their cell, as well as the target label and the probability of the target appearing in the cell. Since the resolution of the grid is significantly reduced compared to the original image, and the detection and recognition steps are processed at the grid cell level, this scheme greatly reduces the computational cost.

[0073] However, because multiple cells use different bounding boxes to predict the same object, this results in many duplicate prediction boxes. This embodiment uses Non-Maximum Suppression (NMS) to address this problem.

[0074] like Figure 5 As shown, the second aspect of the present invention also discloses a tower crane lifting height estimation device according to the target recognition-based tower crane lifting height estimation method of the first aspect, comprising: a controller 3, and a data acquisition module 1; the data acquisition module 1 includes a camera module and a ranging module; wherein,

[0075] The camera module is installed on the tower crane to capture overhead video images of the suspended load and send them to controller 3;

[0076] The ranging module is installed above the hook, specifically at the following location: Figure 2 As shown, the distance from the bottom of the hook to the top of the load, and the height of the hook above the ground are used to obtain the distance. The height of the hook above the ground can be obtained by tower crane sensors, including by sensing and calculating the number of times the hook rope wraps around the drum.

[0077] The controller 3 is used to perform the following steps: receive and store information on the size of the tower base hook and the density of the suspended object; identify the hook boundary and the suspended object boundary in the video image and calculate the pixel ratio of the hook and the suspended object in the video image; calculate the area of ​​the suspended object based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground and the pixel ratio, and calculate the height of the suspended object by combining the weight and density information of the suspended object; calculate the height of the bottom of the suspended object above the ground based on the height of the suspended object, the distance from the bottom of the hook to the top of the suspended object and the hook height above the ground.

[0078] In one embodiment, the camera module employs the following technical means:

[0079] When the camera starts sending a new image frame, a rising edge is generated on the camera's FRM pin. The I / O port connecting the microcontroller and the camera's FRM pin is configured as an external interrupt. When the interrupt occurs, the MCU collects the acquired image information. The camera sends an image frame line by line. For example, for a 120*188 resolution image, the camera divides the image into 120 parts, each containing 188 pixels, and sends them out via an 8-bit parallel port. Whenever the camera starts sending a line of image, a rising edge is generated on the camera's LINE pin. The I / O port connecting the microcontroller and the camera's LINE pin is configured as an external interrupt. When the interrupt occurs, the MCU collects the acquired image information. DMA transfer is enabled within the interrupt handler, and synchronization is performed between the camera and the microcontroller. Each time the I / O port connecting the microcontroller and the camera detects a rising edge, a DMA transfer is performed until all 188 transfers are completed.

[0080] In one embodiment, the ranging module employs the following technical means:

[0081] The ranging module uses an infrared sensor. The infrared emitting circuit's infrared LED emits infrared light, which, after reflection from an obstacle, is received by the photosensitive receiver in the infrared receiving circuit. This allows the sensor to determine if there is an obstacle ahead. The intensity of the reflected light determines the target's distance. Since the light intensity received by the receiver varies with distance, the reflected light is stronger at closer distances and weaker at farther distances. The infrared ranging module then samples the height information H. sIt is sent to the MCU for processing.

[0082] In one embodiment, the data acquisition module is based on the communication module of the remote data acquisition module platform, enabling it to perform functions such as sending and receiving short messages, voice calls, and data transmission. It can be connected to the remote data acquisition module via an RS232 serial port, and various voice and data communication functions can be implemented through AT command control.

[0083] In one embodiment, a display module 4 is also included. The display module 4 is electrically connected to the controller 3 and is used to display video images, as well as the hook boundary and the object boundary in the video images. The display module's HDMI TYPEA pins 1-9 are the pins actually used for TMDS data transmission, divided into three groups: 0, 1, and 2. Pins 10-12 are the TMDS clock signal, pin 13 is a kind of extended HDMI function, allowing manufacturers to customize HDMI messages, pin 14 is a reserved pin, unused (or it could provide an additional pin for CEC), pins 15-16 are I2C pins, pin 17 is the ground pin, pin 18 is the 5V AC pin, and pin 19 is the Hotplug pin (used to monitor the presence of an HDMI device; if present (Hotplug is high), it outputs a DVI signal).

[0084] In one embodiment, a power supply module 2 is also included. The power supply module 2 provides positive and negative voltages to the controller 3, camera module, infrared ranging module, and display module 4. The power supply module includes a power polarity converter and multiple capacitors. The power polarity converter has six pins: 1GND (ground pin), 2SW (output pin), 3VIN (voltage input pin), 4FB (reference voltage pin), 5EN (enable pin), and 6BST (top switch driver pin). Pin 6 is connected to capacitor C47, and pin 4 is connected to capacitor C49 and then grounded. Pin 3 is connected in series with capacitor C60 and then in parallel with capacitor C59 and grounded. The output of pin 2 is connected in parallel with C55, C56, and C53 and then grounded.

[0085] Power module 2 provides a continuous output current of 3A through two integrated N-channel MOSFETs. An internal synchronous power switch provides high efficiency without the need for external Schottky diodes. The DC 12V is connected in series with diode D6SS34 and a 100K resistor R60 to pin 5EN. Pin 6BST is connected in series with capacitor C47 and then to pin 2SW. Pin 2SW is connected to inductor L5, and the 12V-5V conversion is achieved through filtering by capacitors C55, C56, and C57.

[0086] In this embodiment, all resistors in the device have the same accuracy.

[0087] This embodiment comprehensively applies technologies such as target recognition, contour recognition, infrared ranging, and sensors. Cameras installed at the farthest end of the crane boom or on the luffing trolley acquire images of the suspended object. AI target recognition is used to determine the type of object, and sensors obtain weight information to estimate the object's dimensions. This enables the identification of the object type, planar dimensions, and height of the suspended object, and consequently, the determination of the height of the object's bottom from the ground. This provides more refined and practical data support for the tower crane safety monitoring system and collision avoidance devices, reducing collision risks and improving safety management efficiency.

[0088] The following is a computational example of the present invention in a specific scenario:

[0089] Step 1: Configure the tower crane hook plane length L in the system. g Plane width W g With hook height H j Information, in this embodiment, the planar length L of the hook g =1.2m, hook plane width W g =0.15m, hook height H j =0.6m. The system configures the type of suspended object and its corresponding density information. In this embodiment, the suspended object is a stainless steel profile. Due to the placement of the suspended object, there is a cavity. After conversion, its corresponding density is 370kg / m³. 3 .

[0090] Step 2: Acquire video images using the camera on the amplitude-controlled vehicle, such as... Figure 6 As shown.

[0091] Step 3: Obtain the hook height H above the ground using sensors. g In this embodiment, H g Real-time data is acquired via sensors at the bottom of the hook. Taking the data from 13:24:01 on October 25, 2022 as an example, H... g =11.8m.

[0092] Step 4: Identify the upward movement of the hook and determine the moment when the load is fully lifted based on the sensor information. At this moment, determine whether the weight signal obtained by the sensor has reached its maximum value. The sensor will process the sensed weight signal through the microcontroller and display it on the screen.

[0093] Step 5: Obtain and calculate the distance H from the bottom of the hook to the top of the suspended load using the distance measuring device installed on the hook. s . Specifically, H s The height of the hook is calculated by subtracting the hook height from the data measured by the infrared range sensor installed on the upper part of the hook. In this embodiment, the hook height has been entered into the system in step 1. H j =0.6m, H s=1.6-0.6=1m.

[0094] Step 6: Extract frames and determine image pixel availability. Specifically, this is done based on image sharpness.

[0095] Step 7: Identify the hook boundary and calculate its occupied pixels; obtain the object boundary intersecting with the hook boundary and determine it as the object to be lifted. In this embodiment, the hook boundary identification is as follows: Figure 7 The yellow boundary in the image is calculated to occupy 30*209 pixels. The identified boundary of the simulated hanging object is as follows: Figure 7 As shown by the red boundary in the image, it intersects with the yellow boundary of the hook.

[0096] Step 8: Identify the type of object to be lifted, obtain the corresponding density from the database, identify the boundary of the object, and calculate the number of pixels it occupies. In this embodiment, a target recognition algorithm is used to identify the type of object to be lifted. The object to be lifted in this case is identified as a stainless steel profile with a corresponding density of 370 kg / m³. 3 Next, identify the boundaries of the object to be lifted, such as... Figure 7 As shown by the red boundary, the calculated area of ​​the object to be suspended is 355*239 pixels.

[0097] Step 9: When the boundary of the object to be suspended exceeds the current camera's outer frame, reduce the camera's focal length and take a new screenshot.

[0098] Step 10: Calculate the area of ​​the object to be suspended based on the hook's height from the ground, the hook area, and the relationship between the pixels occupied by the hook and the pixels occupied by the object to be suspended. Specifically, due to the height difference between the object and the hook, there is a certain error between the ratio of the image area of ​​the object and the hook and the actual area ratio. This error ratio can be approximated as the square of the ratio of the distance between the camera and the object to the distance between the camera and the hook. Therefore, the area of ​​the object to be suspended is equal to the product of the ratio of the number of pixels recognized in the object image to the number of pixels recognized in the hook image, multiplied by the hook area, and then multiplied by the perspective error ratio.

[0099] In this embodiment, Figure 7 Taking the data from 13:24:01 on October 25, 2022 as an example, according to the real-time data fed back by the sensor, the distance H of the boom above the ground is 60m, and the distance H of the hook above the ground is... g =11.8m, the distance between the camera and the suspended object is equal to the distance H of the boom from the ground plus the height H of the hook from the top surface of the suspended object. s Subtract the distance H of the hook from the ground. g That is, H+H s -H g = 49.2m; The distance between the camera and the hook is the distance H from the boom to the ground minus the hook height H. j Subtract the distance H of the hook from the ground. g HH j -Hg = 47.6m; therefore, the area of ​​the object to be lifted is:

[0100] S=[1.2*0.15*(355*239) / (30*209)] / (47.6 / 49.2) 2 ≈2.60m 2 .

[0101] Step 11: Obtain the weight of the object to be lifted, and calculate the lifting height H based on the density and area of ​​the object. w Specifically, the weight of the suspended object is obtained by the weight sensor built into the tower crane. In this embodiment, the weight of the suspended object is 850 kg, the density of the suspended object is the data pre-entered into the system in step 1, and the area of ​​the suspended object is obtained from step 9. Since the shape of the suspended object in this embodiment is a cuboid, the height H of the suspended object is... w H is calculated by dividing the weight of the suspended object by its area and then by its density. w =850 / (370*2.60)≈0.88m.

[0102] Step 12: Determine the height H of the suspended load above the ground. d . Specifically, H d Equal to the height H of the bottom of the hook from the ground g Subtract the distance H from the bottom of the hook to the top of the load. s Subtract the height H of the suspended object. w H g The height is obtained in real time by the height sensor at the bottom of the hook, H s H is calculated from step 5. w Calculated from step 10. In this embodiment, H d =11.8-1-0.88=9.92(m).

[0103] Actual distance: Measured by sensors, at 13:24:01 on October 25, 2022, H g =11.8m, Hs=1m, the weight of the suspended object is 850Kg. The actual measured length of the suspended object is 2.38m, and the width is 1.60m, with an error of 1.21m between this measurement and the actual area. 2 The actual measured height of the suspended object was approximately 1.4m, and the height of the object from the ground was: Hd = 11.8 - 1 - 1.4 = 9.4m, with an error of 0.52m compared to the height calculated by this method. The main sources of error between this method and the actual test are: an underestimation of the area due to the target recognition algorithm error, and an error caused by excessive density due to gaps between the suspended objects. These two errors are coupled. The error due to underestimation of the area is 0.6 - 1.4 = (-0.8)m, and the error due to the density of the suspended object is: -0.8 + 0.52 = 0.28m. The resulting error range is (-0.8, 0.28).

[0104] The above provides a detailed description of the tower crane lifting height estimation method and device based on target recognition provided by the present invention. Specific examples are used in this embodiment to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

[0105] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined in these embodiments may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for estimating the height of a tower crane load based on target recognition, characterized in that, Includes the following steps: Obtain information on the dimensions of the tower base hook and the density of the suspended load; After obtaining the weight information of the suspended object and determining that the object has been lifted, obtain the distance from the bottom of the hook to the top of the suspended object, as well as the height of the hook above the ground; Acquire overhead video images of the suspended object, the video images containing complete images of the hook and the suspended object; identify the boundaries of the hook and the suspended object in the video images, calculate the pixel ratio of the hook and the suspended object in the video images, and determine the type of the suspended object; The area of ​​the suspended object is calculated based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground, and the pixel ratio. The height of the suspended object is calculated by combining the weight and density information of the suspended object. The steps for calculating the area of ​​a suspended object include: ; Among them, S d The top view area of ​​the hook is calculated based on the hook dimensions, where H is the distance from the boom to the ground. j H is the height of the hook. g H is the height of the hook from the ground. s S' / S is the distance from the bottom of the hook to the top of the load. d 'This represents the pixel ratio of the hook and the suspended load in the video image; Calculate the height of the bottom of the load from the ground based on the height of the load, the distance from the bottom of the hook to the top of the load, and the height of the hook from the ground.

2. The tower crane lifting height estimation method based on target recognition according to claim 1, characterized in that, The dimensions of the tower base hook include hook length, hook width, and hook height.

3. The tower crane lifting height estimation method based on target recognition according to claim 1, characterized in that, The steps to obtain the density information of the suspended object include: Pre-store different types of suspended objects and the corresponding density information for each type of suspended object; Automatically obtain the density information of the suspended object based on the current type of suspended object.

4. The tower crane lifting height estimation method based on target recognition according to claim 1, characterized in that, The steps for calculating the height of the bottom of the load from the ground, based on the height of the load, the distance from the bottom of the hook to the top of the load, and the height of the hook above the ground, include: ; Among them, H g H is the height of the hook from the ground. s M is the distance from the bottom of the hook to the top of the load, ρ is the weight of the load, and S is the density of the load. d 'S' represents the pixel area of ​​the hook in the video image. d The top view area of ​​the hook is calculated based on the hook size information.

5. The tower crane lifting height estimation method based on target recognition according to claim 1, characterized in that, The steps for identifying the hook boundary and the suspended object boundary in the video image using the target recognition algorithm include: Each frame of video image is divided into N grid units. The boundaries of the target object to be lifted and the hook boundary within the grid are identified. The type of the object to be lifted is determined. The corresponding type density is obtained from the database. The pixels occupied are calculated to realize the detection and positioning of the target object to be lifted and the hook. Predict the relative coordinates of the target with respect to the bounding box of its grid cell, as well as the target label and the probability that the target appears in the cell.

6. A tower crane lifting height estimation device according to any one of claims 1-5 based on the target recognition-based tower crane lifting height estimation method, characterized in that, include: The controller, camera module, and ranging module; among them, The camera module is installed on the tower crane and is used to capture overhead video images of the suspended load and send them to the controller. The ranging module is installed at different positions on the hook to obtain the distance from the bottom of the hook to the top of the suspended object, as well as the height of the hook above the ground; The controller is used to perform the following steps: receive and store information on the size of the tower base hook and the density of the suspended object; identify the hook boundary and the suspended object boundary in the video image, and calculate the pixel ratio of the hook and the suspended object in the video image; calculate the area of ​​the suspended object based on the hook size information, the distance from the bottom of the hook to the top of the suspended object, the hook height above the ground, and the pixel ratio, and calculate the height of the suspended object based on the weight and density information of the suspended object; calculate the height of the bottom of the suspended object above the ground based on the height of the suspended object, the distance from the bottom of the hook to the top of the suspended object, and the hook height above the ground.

7. The tower crane lifting height estimation device according to claim 6, characterized in that, Also includes: The display module, which is electrically connected to the controller, is used to display video images, as well as the boundaries of the hook and the suspended object in the video images.

8. The tower crane lifting height estimation device according to claim 6, characterized in that, Also includes: A power module, which is electrically connected to the controller, is used to supply power to the controller, camera module, and ranging module.

9. The tower crane lifting height estimation device according to claim 6, characterized in that, The ranging module uses an infrared sensor.

Images