CN112027842A - Method for detecting brake wheel corner in braking process of elevator brake - Google Patents

Method for detecting brake wheel corner in braking process of elevator brake Download PDF

Info

Publication number
CN112027842A
CN112027842A CN202011076670.3A CN202011076670A CN112027842A CN 112027842 A CN112027842 A CN 112027842A CN 202011076670 A CN202011076670 A CN 202011076670A CN 112027842 A CN112027842 A CN 112027842A
Authority
CN
China
Prior art keywords
image
brake
binary image
edge
brake wheel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011076670.3A
Other languages
Chinese (zh)
Inventor
冯双昌
欧阳惠卿
常晓清
陈杰
方良
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Special Equipment Supervision and Inspection Technology Institute
Original Assignee
Shanghai Special Equipment Supervision and Inspection Technology Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Special Equipment Supervision and Inspection Technology Institute filed Critical Shanghai Special Equipment Supervision and Inspection Technology Institute
Priority to CN202011076670.3A priority Critical patent/CN112027842A/en
Publication of CN112027842A publication Critical patent/CN112027842A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
    • B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
    • B66B5/0006—Monitoring devices or performance analysers
    • B66B5/0037—Performance analysers
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66D—CAPSTANS; WINCHES; TACKLES, e.g. PULLEY BLOCKS; HOISTS
    • B66D5/00—Braking or detent devices characterised by application to lifting or hoisting gear, e.g. for controlling the lowering of loads
    • B66D5/02—Crane, lift hoist, or winch brakes operating on drums, barrels, or ropes
    • B66D5/06—Crane, lift hoist, or winch brakes operating on drums, barrels, or ropes with radial effect
    • B66D5/08—Crane, lift hoist, or winch brakes operating on drums, barrels, or ropes with radial effect embodying blocks or shoes
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66D—CAPSTANS; WINCHES; TACKLES, e.g. PULLEY BLOCKS; HOISTS
    • B66D5/00—Braking or detent devices characterised by application to lifting or hoisting gear, e.g. for controlling the lowering of loads
    • B66D5/02—Crane, lift hoist, or winch brakes operating on drums, barrels, or ropes
    • B66D5/24—Operating devices
    • B66D5/30—Operating devices electrical
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/02—Affine transformations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G06T5/30—Erosion or dilatation, e.g. thinning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66D—CAPSTANS; WINCHES; TACKLES, e.g. PULLEY BLOCKS; HOISTS
    • B66D2700/00—Capstans, winches or hoists
    • B66D2700/03—Mechanisms with latches or braking devices in general for capstans, hoists or similar devices as well as braking devices actuated electrically or by fluid under pressure
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Mechanical Engineering (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a method for detecting a brake wheel corner in the braking process of an elevator brake, which is characterized by comprising the following steps of: a plurality of equidistant marks are arranged at positions close to the edges on the brake wheel, an industrial camera is used for shooting the braking process of the brake wheel in the elevator brake while the elevator brake brakes, the shot video image data consisting of n frames of images are transmitted to a computer, and the computer calculates and displays the corner of the brake wheel by utilizing an image processing algorithm. The invention realizes non-contact measurement, the performance of the brake cannot be influenced by the measuring device, destructive modification of the brake is not needed, the installation is convenient, the parameters can be automatically updated during each calculation, the influence of environment change on the detection system is reduced, and the long-term stable work of the detection system is facilitated.

Description

Method for detecting brake wheel corner in braking process of elevator brake
Technical Field
The invention relates to a method for detecting a brake wheel corner of an elevator traction braking device, and belongs to the technical field of elevator detection.
Background
The elevator mainly comprises a traction braking device, a control cabinet, a car, a counterweight and the like. One end of the steel wire rope is connected with the lift car and wound on a traction wheel of the traction braking device, and the other end of the steel wire rope is connected with a counterweight for balancing. The car is used for carrying passengers; the counterweight is used for balancing the weight of the lift car, so that unnecessary work is reduced; the traction braking device controls the rotation and the stop of the traction sheave and drives the lift car to ascend or descend by utilizing the friction force between the traction sheave and the steel wire rope. The elevator traction system and the brake system are two major core safety protection systems of an elevator, wherein the brake system ensures the reliable braking of the elevator in the running process, and the brake system fails, so that the elevator cannot be braked in time, and an elevator car top-rushing or bottom-rushing accident occurs. Therefore, a system capable of detecting the braking capability of the elevator is needed, and relevant parameters of the braking process of the elevator at each time are detected, recorded and displayed, so that a maintainer can conveniently know the braking capability of the elevator in time and overhaul the elevator. An effective brake detection system is of great significance to prevent casualties and reduce economic loss.
The elevator brake comprises a pair of brake arms, one end of each brake arm is provided with an electromagnet and a spring, the other end of each brake arm is hinged with a brake base, a brake shoe is arranged in the middle of each brake arm and is opposite to a brake wheel, the springs extend under the power-off state of the brake, the brake shoes are pressed on the brake wheels by the springs, and the brake wheels are stopped by means of the friction force of the brake shoes and the brake wheels, so that the brake effect is achieved. When the brake is used for a long time, the brake shoe is gradually worn, so that the spring needs to be stretched longer in the compression process and is closer to the original length of the spring, the compression force of the spring is reduced, the friction force between the brake wheel and the brake shoe is reduced, the braking force is insufficient, and the brake fails. There are many ways for an elevator brake monitoring system to monitor the braking capacity of the brake. The existing detection device mainly detects parameters such as the suction force of the brake electromagnet, coil current, braking and stopping distance and the like, the detection of the electromagnetic suction force can only be positioned on a special test bed, and the detection device is not suitable for field detection and real-time monitoring. The coil current detection is convenient, but the braking performance of one brake cannot be directly reflected. The pulley and the encoder are needed to be used for measuring the braking distance, the principle is simple, the installation is complex, and the conditions of reduced installation stress, pulley slippage and the like need to be considered when the device is used for a long time. If only the braking capability of the braking wheel is detected, the rotation angle of the braking wheel in the braking process is a more ideal index, and the quality of the braking capability of one brake can be directly reflected
The invention patent application with the application number of 201911096218.0 and the publication number of CN110790103A discloses a testing device and a testing method, which relate to the testing technology of an electromagnetic brake and are used for testing the braking performance of the electromagnetic brake. The testing device and the testing method provided by the invention can replace manual work to test the braking performance of the electromagnetic brake, such as the braking force, the electromagnetic suction force in a voltage reduction power supply state, the braking balance degree, the free clearance and the like. The testing technology needs to build a test bed for testing, and is not suitable for field testing and real-time monitoring.
Disclosure of Invention
The purpose of the invention is: and processing the input image data by using an image algorithm and a high-frame camera so as to obtain the total rotation angle of the elevator brake in the braking process.
In order to achieve the aim, the technical scheme of the invention provides a method for detecting the corner of a brake wheel in the braking process of an elevator brake, which is characterized by comprising the following steps of:
the method comprises the following steps that a plurality of equidistant marks are arranged at positions close to the edges on a brake wheel, an industrial camera is used for shooting the braking process of the brake wheel in the elevator brake while the elevator brake brakes, the shot video image data consisting of n frames of images are transmitted into a computer, and the computer calculates and displays the corner of the brake wheel by using an image processing algorithm, wherein the image processing algorithm comprises the following steps:
the method comprises the following steps of 1, extracting coordinates of edge points of the outer contour of the brake wheel of any frame of image in a video image, and fitting by a least square method to obtain a parameter equation which is met by the contour of the brake wheel in an image coordinate system u-v;
step 2, reversely solving an external parameter matrix T' of the industrial camera according to the parameter equation obtained in the step 1;
step 3, identifying the marker of each frame of image in the video image to obtain the corner of two adjacent frames of images, and setting the corner of the kth frame of image and the (k + 1) th frame of image as deltak,k+1Then angle of rotation Δk,k+1Comprises the following steps:
step 301, detecting the markers of the kth frame image to obtain a set of coordinates of all the markers in an image coordinate system u-v, wherein the coordinates of the ith marker in the kth frame image in the image coordinate system u-v are (u-v)i,vi)；
Step 302, utilizing an inverse matrix T 'of the extrinsic matrix T'-1Transforming the coordinates of all the markers in the k frame image under the image coordinate system u-v into a world coordinate system xw-yw-zwCoordinates of the ith marker in the kth frame image in the world coordinate system xw-yw-zwThe coordinates of (x) belowi,yi)；
Step 303, utilizing the world coordinate system x of each marker in the k frame imagew-yw-zwCalculating the coordinate to obtain the included angle of each marker, and setting the included angle of the ith marker in the kth frame image as thetak,iThen, a set S of all included angles of the k frame image is obtainedk＝{θk,i:i＝1,2,…}；
Step 304, obtaining a set S of all included angles of the k +1 frame image by adopting the same method from the step 301 to the step 303k+1＝{θk+1,i:i＝1,2,…}；
Step 305, utilizing set Sk＝{θk,iI 1,2, … and set Sk+1＝{θk+1,iI-1, 2, … construct a new set Ak,k+1，Ak,k+1＝{|θk,i-θk+1,j|:θk,i∈Sk,θk+1,j∈Sk+1}；
Step 306, from set Ak,k+1To find out all of
Figure BDA0002717023210000031
The values of (1) are averaged to obtain a rotation angle delta from the k frame image to the (k + 1) th frame imagek,k+1；
And 4, summing the corners of all the two adjacent frames of images to obtain the corner of the brake wheel.
Preferably, the step 1 comprises the steps of:
step 101, after obtaining a gradient map f (u, v) of any frame of image I in the video image, taking a threshold value T for the gradient map f (u, v)min<f(u,v)<TmaxTo obtain a binary image Bgrad；
Step 102, for the binary image BgradScreening the points to eliminate points not belonging to the edge of the brake wheel to obtain a binary image B with comprehensively filtered color and gradient thresholds1；
Step 103, binary image B1Contains two edge lines: the gradient direction ranges on two edge lines are different, and the required outer edge is reserved through gradient direction filtering to obtain a binary image B after the gradient direction filtering2；
Step 104, removing the binary image B2Obtaining a binary image B from non-edge detailsrm；
Step 105, the binary image B is processed by the following formularmPerforming edge restoration to obtain a binary image Xfinal：
Figure BDA0002717023210000032
In the formula, X0Representing a binary image obtained by the 0 th iteration; xiRepresenting a binary image obtained by the ith iteration; s is a convolution kernel of image expansion; iterate to XfinalIn the process, the edge can be well restored without introducing unnecessary non-edge details again, and the final value obtains the best effect through actual test;
step 106, for the binary image XfinalEdge thinning is carried out to obtain a single-pixel edge binary image BcannyBy plotting a binary image XfinalAnd a single-pixel edge binary image BcannyObtaining a refined edge binary image B by intersectionedge；
Step 107, obtaining a through-edge binary image B by least square fittingedgeThe obtained contour of the brake wheel satisfies a parametric equation in the image coordinate system u-v.
Preferably, in the step 102, the binary image B is processedgradWhen the points in the image are screened, a binary image B is setgradThe coordinate of a certain point A is (u)0,v0) In the image I (u, v) by (u)0,v0) In the region of n × n pixels at the center, if the color value in HLS color space is satisfiedmin<h<hmaxAnd smin<s<smaxThe number of points in the range being greater than nminAnd is less than nmaxThe point A is retained, otherwise from the binary image BgradPoint a is removed.
Preferably, in the step 103, the binary image B is taken from the image coordinate system u-v1Obtaining the binary image B by the points with the medium gradient direction angle in the range of (-arctank, arctank)2K is the angle between the gradient directionsThe slope corresponding to half of the angle of (1) satisfies:
Figure BDA0002717023210000041
in the formula, graduIs a binary image B1The gradient of any point in the u direction; gradvIs a binary image B1The gradient of any point in the v direction.
Preferably, the optical axis of the camera forms an included angle alpha with the rotating shaft of the braking wheel in a transverse plane, and the optical axis of the camera forms an included angle beta with the rotating shaft of the braking wheel in a longitudinal plane, wherein alpha is more than or equal to-20 degrees and less than or equal to 20 degrees, and beta is more than or equal to-20 degrees and less than or equal to 20 degrees.
Preferably, the space distance from the center of the lens of the camera to the center of the front end face of the elevator brake is d, and d is more than or equal to 800mm and less than or equal to 1200 mm.
The invention uses image processing, adds parameter correction algorithm in the image algorithm, and corrects the parameters once before each calculation, so that the installation condition of the camera is simplified, and the camera can work reliably for a long time. By using image recognition, non-contact measurement can be realized, the normal use of the elevator can not be influenced in the measuring process, the brake is not required to be destructively transformed, and the requirement of real-time monitoring can be met.
The invention realizes non-contact measurement, the performance of the brake cannot be influenced by the measuring device, destructive modification of the brake is not needed, the installation is convenient, the parameters can be automatically updated during each calculation, the influence of environment change on the detection system is reduced, and the long-term stable work of the detection system is facilitated.
Drawings
FIG. 1 is a schematic diagram of the system of the present invention;
FIG. 2 is a parameter diagram of the present invention;
FIG. 3 is a schematic diagram of a coordinate system of the present invention;
FIGS. 4a to 4f are graphs B of the two-valued edge obtained according to the present inventionedgeA schematic diagram of (a);
FIGS. 5a and 5b are schematic views of the fitted edge of the brake wheel;
FIG. 6 is a top view of a camera mounting angle;
FIG. 7 is a side view of a camera mounting angle;
fig. 8 is a schematic view of the distance between the camera and the stopper.
Detailed Description
The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
The invention provides an elevator brake rotation angle detection method based on image recognition, and as shown in figure 1, the invention adopts high frame rate video collected by an industrial camera as input. Because the rotating speed of the brake wheel is high in the running process, an industrial camera with high frame rate and short exposure time is required to be used for completely acquiring the motion state of the brake wheel in one rotation. In addition, in order to reduce the strict requirement on the relative position of the industrial camera and the brake wheel, the calculation parameters need to be updated according to the acquired data, and when the relative position of the industrial camera and the brake wheel slightly changes, the industrial camera can still accurately calculate the rotation angle. The high frame rate video collected by the industrial camera is forwarded to the computer through the router, the computer processes the received data through an image processing algorithm, the detection of the rotation angle of the elevator brake is realized, and the detection result is displayed by the monitor.
When the camera is placed, the center of the visual field of the lens is aligned with the center of the front end face (the circular face where the identification edge of the brake wheel is located) of the brake wheel, so that the optical axis of the camera is aligned with the center of the brake wheel (in actual conditions, the camera does not need to be aligned with the center of the brake wheel strictly, and only needs to be approximate). Thus, the camera adds two constraints, leaving 4 degrees of freedom, three attitude angles and one distance, respectively.
The three attitude angles can be represented as an angle α between the optical axis of the camera 1 and the rotation axis of the headblock 2 in the horizontal plane as shown in fig. 6, an angle β between the optical axis of the camera 1 and the rotation axis of the headblock 2 in the longitudinal plane as shown in fig. 7, and an angle γ by which the camera rotates along its own optical axis, wherein the angles α and β are not fixed values, and an appropriate angle can be selected within a given range according to actual conditions. Wherein the alpha angle is more than or equal to-20 degrees and less than or equal to 20 degrees, and the beta angle is more than or equal to-20 degrees and less than or equal to 20 degrees.
The angle γ can be chosen arbitrarily, and for convenience of image processing, the camera 1 can be rotated along the optical axis to make the unshielded part of the brake wheel 2 fall on the right side of the image as much as possible.
As shown in fig. 8, the distance d from the camera 1 to the stopper 3 refers to the spatial distance d from the lens center of the camera 1 to the center of the front end face of the stopper 3, and the distance d is related to the selection of the camera 1, under the condition that the frame rate (above 200 fps), the view range (covering the whole brake wheel) and the resolution (being capable of clearly identifying the mark) of the camera meet the requirements. The distance required by the camera for clear imaging is d, and the range of d is generally 800 mm-1200 mm.
The invention gives the basic parameters of the camera fixing position and the brake wheel, as shown in fig. 2, wherein the parameters are explained as follows:
d is the diameter of the brake wheel;
r-the braked wheel radius;
R0-distance of the camera lens center to the headblock center;
alpha is the included angle of the projection of the center of the camera lens, the center of the brake wheel and the axis of the brake wheel on the horizontal plane.
The invention provides an integral scheme of a method for detecting the corner of a brake wheel in the braking process of an elevator brake, which is specifically described as follows:
after the control cabinet of the elevator brake sends out a braking signal, the braking of the elevator brake is controlled and simultaneously input into the computer to inform the computer that the braking is started. And the computer sends a control signal after receiving the braking signal, and sends the control signal to the industrial camera through the industrial router to control the industrial camera to acquire images. The industrial camera acquires image information in the elevator braking process and sends the image information to the computer through the industrial routerThe machine is cached by the computer and starts timing when the maximum possible braking time t of the elevator is exceededmaxAnd when the industrial camera is in use, the computer sends out a control signal to control the industrial camera to stop collecting images.
And (3) processing the images by using an image recognition algorithm while caching the images, wherein the image recognition algorithm firstly extracts the outline of the edge of the brake wheel in the image and fits the outline by using an elliptic equation. And calculating and solving the external parameters of the camera according to the obtained elliptic equation, the real size of the brake wheel and the internal parameters of the camera obtained by calibration. And finally, inversely transforming the identified marker in the graph back to the world coordinate according to the camera external parameter matrix, and solving the current rotation angle of the brake wheel according to the change of the world coordinate.
According to the foregoing description, the basic steps of the image processing algorithm are:
firstly, extracting and fitting the edge contour of the brake wheel;
secondly, solving a current external parameter matrix of the camera;
and thirdly, calculating a rotation angle.
The image processing algorithm is described in detail as follows:
(1) and extracting and fitting the edge profile of the brake wheel.
The method comprises the steps of extracting and fitting the contour of the brake wheel, namely extracting coordinates of edge points of the outer contour of the brake wheel through image processing, and obtaining an equation of the contour of the brake wheel in an image coordinate system through least square fitting. The process comprises three steps: extracting the contour, establishing a parameter equation and fitting by a least square method.
Gradient threshold filtering is a common method of edge feature extraction. Images tend to have large gray scale gradients at the edges. By this feature, points at the edge in the image can be preliminarily extracted, including the edge points of the brake wheel. Image gradient calculation usually adopts a sobel operator, firstly, an image is converted into a gray image, Gaussian blur is carried out on the gray image, gradient calculation in the u direction and the v direction is respectively carried out, then gradients in the u direction and the v direction are combined into a gradient vector, and the modulus is that:
Figure BDA0002717023210000071
in formula (1), I represents an arbitrary frame image acquired by an industrial camera; mag (I) is the modulus of the gradient vector of any point; grad (I) is the gradient vector of any point; gradu(I) The gradient of any point in the gray scale image I in the u direction is shown; gradv(I) The gradient of any point in the gray scale image I in the v direction is shown.
After obtaining the modulus of the gradient of the image at any point (u, v). Normalizing the value range of the standard to be in the range of [0,255] to facilitate subsequent screening:
Figure BDA0002717023210000072
in the formula (2), f (u, v) is a gradient diagram; [. is a Gaussian function (rounded down); mag (I (u, v)) ═ mag (I)) is the modulus of the gradient vector at any point.
Taking a threshold T for the function f (u, v)min<f(u,v)<TmaxTo obtain a binary image BgradBinary image BgradContains many points that do not belong to the edge of the headblock. For this binary image BgradFurther screening the points in (A) to obtain a binary image BgradWhere the color of each point around the same coordinate of the image I (u, v) is counted. For the binary map BgradThe coordinate of a certain point A is (u)0,v0) In the image I (u, v) by (u)0,v0) In the region of n × n pixels at the center, if the color value in HLS color space is satisfiedmin<h<hmaxAnd smin<s<smaxThe number of points in the range being greater than nminAnd is less than nmaxThe point A is retained, otherwise from the binary image BgradPoint a is removed. Traversing the binary image B by adopting the methodgradObtaining a new binary image B after all the points in the image1. Binary image B1Namely the result of the comprehensive filtering of the color and the gradient threshold. Wherein n, hmin、hmax、smin、smax、nmin、nmaxThe selection of the parameters is related to camera parameters, brakes and field conditions, and the parameters are obtained by testing according to actual conditions.
Binary image B1Contains two edge lines: the inner edge and the outer edge of the stopper are different in the direction range of the gradient on the measuring edge line, and the required outer edge is reserved through gradient direction filtering. In the u-v plane, taking a binary image B1And the medium gradient direction angle is at a point in a range (-arctank, arctank), and k is a slope corresponding to a half angle of an angle included by the gradient direction range, namely, the following conditions are satisfied:
Figure BDA0002717023210000081
in the formula (3), graduIs a binary image B1The gradient of any point in the u direction; gradvIs a binary image B1The gradient of any point in the v direction.
Obtaining a new binary image B2I.e. the result obtained after the gradient direction filtration.
Binary image B2There remain tiny interference points that do not belong to the edge and an algorithm can be used to remove these details. The method that can be adopted is: statistical binary image B2A neighborhood of all non-zero points in the neighborhood (e.g., a 13 × 13 square region centered on the point, the current point is defined as the center point), the number of non-zero points in the neighborhood is calculated, and when the number is greater than a certain threshold (e.g., 30), the center point is retained. When less than the threshold, this center point is removed (assigned a value of 0), so that isolated local details can be removed, resulting in a binary image Brm。
Binary image BrmThe binary image B is removed2The non-edge details in (1) also remove a part of the edge points, and the edge restoration needs to be performed by the following formula:
Figure BDA0002717023210000091
in the formula (4), X0Representing a binary image obtained by the 0 th iteration; xiRepresenting a binary image obtained by the ith iteration; s is a convolution kernel for image expansion, and can be selected as a square area of 3 × 3 or 5 × 5 as required.
Iterate to XfinalThe edges can be restored better without reintroducing unwanted non-edge details. The final value can be best tested by actual testing.
Binary image XfinalThe edges in (1) are thick, which is not beneficial for calculation and needs edge thinning. Applying Canny edge detection algorithm to binary image XfinalProcessing to obtain a single-pixel edge binary image B of the whole imagecannyFinally, by mixing XfinalAnd a single-pixel edge binary image BcannyObtaining a refined edge binary image B by intersectionedge。
(2) Solving a current external parameter matrix of an industrial video camera (hereinafter referred to as a camera).
The external reference matrix is required to be solved, the brake wheel edge contour fitting needs to be carried out firstly, and the parameter equation which is met by the brake wheel contour in the image coordinate system u-v needs to be obtained firstly when the contour fitting is carried out. Given the contour of the brake wheel in the world coordinate system xw-yw-zwThe next is a circle. From world coordinates to image coordinates, two transformations need to be undergone: a perspective transformation and an affine transformation (without taking into account camera distortion, since the distortion has been eliminated by camera calibration). The formula of the perspective transformation is as follows:
Figure BDA0002717023210000092
in the formula (5), xc、yc、zcThe coordinate of a certain point in the space in a camera coordinate system; x is the number ofw、yw、zwThe coordinate of a certain point in the space in a world coordinate system;
Figure BDA0002717023210000093
for a transformation matrix from camera coordinates to world coordinates, rij、tiThe parameters of the matrix T are the camera external parameters, i is 1,2,3, j is 1,2, 3; and p and q are p-q plane coordinates, and the p-q plane is a plane of the original world coordinate plane after perspective transformation.
The formula of the camera affine transformation is:
Figure BDA0002717023210000101
in the formula (6), fx、fy、cx、cyThe affine transformation parameter is only related to the parameter of the camera, and is not related to the position of the camera, and is called as internal reference.
The equation (circular equation) of the contour of the brake wheel under the world coordinate needs to be converted for two times to obtain an equation in a u-v plane, so that a parameter equation becomes very complex. Since the camera internal parameters can be determined during camera calibration and will not change with the change of the camera position, it is possible to make:
Figure BDA0002717023210000102
through the formula (7), the contour point set extracted in the u-v plane can be determined (u-v plane)i,vi) The transformation is set of coordinate points (p)i,qi) The computational effort can be reduced by 1,2, n, n being the total number of contour points extracted in the u-v plane and then calculating the equations satisfied in the p-q plane.
Setting the relation formula of the contour of the brake wheel under the world coordinate as follows:
Figure BDA0002717023210000103
in equation (8), θ is the current attitude angle of the brake wheel, and any attitude can be used as a reference.
The following equations (5) and (8) can be used to find that p and q satisfy the following relations:
Figure BDA0002717023210000104
in the formula (9), kijIs composed of
Figure BDA0002717023210000105
i＝1,2,3，i＝1,2；siIs composed of
Figure BDA0002717023210000106
Transforming equation (9) to obtain:
Figure BDA0002717023210000111
the two formulas of the formula (10) are combined, and theta is eliminated to obtain the shape like Ap2+Bq2The equation of + Cpq + Dp + Eq ═ 1, which is exactly one quadratic curve, and the closed figure in the quadratic curve is only an ellipse, so the equation must be an ellipse. The ellipse can be fitted using a least squares fit to find A, B, C, D, E five parameters.
According to the above results { (p)i,qi) One can get a set of contradictory equations:
Figure BDA0002717023210000112
i.e., Ax ═ b form, the least squares solution of this equation is x ═ a+b, thus obtaining all parameters.
Five parameters can be obtained A, B, C, D, E by contour fitting. These parameters are used to solve the camera external parameter matrix (i.e. the matrix T in equation (5)) which contains 12 parameters in total, but not all of them, where all r are free parametersijAnd forming a rotation matrix which is an unit orthogonal matrix and has only three variables. Thus, the matrix T can be written as:
Figure BDA0002717023210000113
in the formula (12), Rx(α)、Ry(β)、RzAnd (γ) represents a rotation matrix of the coordinate system by rotation angles α, β, γ around the x, y, z axes thereof, respectively, where γ is 0.γ is 0 because, in world coordinates, the contour of the headblock is an ellipse, regardless of whether the contour is along zwAnd how to rotate, and finally, the same ellipse equation can be obtained in the image coordinate system. Therefore, the parameter γ can be arbitrarily taken. Substituting the parameters in the matrix into r in the following formula (13)ij：
Figure BDA0002717023210000114
The resulting p, q are then related to α, β, γ and t1、t2、t3Substituting the expression of (2) into Ap2+Bq2+ Cpq + Dp + Eq ═ 1, a new quadratic equation A' x is obtainedw 2+B′yw 2+C′xwyw+D′xw+E′ywF ', the quadratic curve is the equation of the contour of the brake wheel in the world coordinate system, i.e. the equation of a circle with a radius R, where a ', B ', C ', D ', E ', F ' satisfy:
Figure BDA0002717023210000121
only 5 equations can be listed by 5 parameters of the ellipse, a group of parameters which are in accordance with the equation (14) is solved, and the general solution of gamma under any value can be obtained by rotation transformation.
Solve α, β, t inversely by equation (14)1、t2、t3A non-linear system of equations needs to be solved, which has no formula solution. But the parameters can be approximated by a gradient descent method to obtain an approximate solution. Constructing a loss function:
En＝(F′-R2A′)2+(F′-R2B′)2+C′2+D′2+E′2 (15)
and performing gradient reduction by utilizing an rmsprop gradient reduction algorithm until En is lower than a given threshold Tn, so that a satisfactory result can be obtained according to actual condition tests. Calculating alpha, beta, t1、t2、t3After the approximation of (3), the extrinsic parameter matrix T' can be obtained.
(3) And (5) calculating the rotation angle.
The angle calculation is mainly divided into two steps: and identifying the coordinates of the markers on the brake wheel and calculating the rotation angle.
When the angle is calculated, marks with equal intervals are firstly marked on the brake wheel close to the edge, some brake wheels are provided with identifiable marks, and some marks can be stuck to the edge of the brake wheel for brake wheels without the marks.
The simplest method for identifying the marker is to use a marker with a special color, and locate the coordinates of the marker in the graph through color identification. Because the braked wheel is not completely positioned in the camera view, partial shielding can occur, a plurality of markers need to be set, the determination of the distance between the markers needs to be carried out according to the rotating speed of the braked wheel and the frame number of the camera, and in order to meet the sampling theorem, the camera needs to shoot more than two frames at least in the time interval when the next marker reaches the original position of the previous marker. Generally, four marks are set, and every two marks are separated by
Figure BDA0002717023210000131
A circular arc.
After obtaining the coordinates of the markers in the image, first, the inverse matrix T ' of the external reference matrix T ' and equation (7) are used '-1Coordinates (u) of all markers in the graph under the image coordinate system u-vi,vi) Conversion to world coordinate system xw-yw-zwCoordinates of lower (x)i,yi) And then according to:
Figure BDA0002717023210000132
and calculating to obtain the current frame included angle theta (the current frame is taken as the kth frame). Since more than one marker may be detected per frame, a set S of angles is obtainedk＝{θk,iI 1,2, … }. Similarly, for the (k + 1) th frame, another set S is obtainedk+1＝{θk+1,iI 1,2, … }, a new set is constructed:
Ak,k+1＝{|θk,i-θk+1,j|:θk,i∈Sk,θk+1,j∈Sk+1} (18)
because the markers are spaced apart
Figure BDA0002717023210000133
Arc, the angle of the brake wheel of two adjacent frames must be less than
Figure BDA0002717023210000134
Circular arcs, from the set Ak,k+1To find out all of
Figure BDA0002717023210000135
The value of (c). These angles are approximately equal to the rotation angle of two adjacent frames, and the filtered angles are averaged to obtain the rotation angle delta from the k frame to the k +1 framek,k+1Assuming that the braking process has n frames of images in total, the total rotation angle is:
Figure BDA0002717023210000136
the description of the embodiment of the present invention will be made by taking the brake process rotation angle detection of the drum brake shown in fig. 4a as an example.
The key point of the invention is that a brake wheel corner detection method in the elevator braking process is designed aiming at an elevator brake. The system block diagram on which the method relies is shown in fig. 1. When the elevator brake control cabinet gives a signal to stop the computer by using the industrial camera which is installed in advance, the computer starts to receive the image input of the industrial camera, and the image acquired by the industrial camera needs to be forwarded to the computer once. The computer completes all image processing and calculation tasks, and the display screen displays calculation results.
Aiming at the overall scheme, the invention provides the following image processing specific steps:
firstly, extracting and fitting the edge contour of the brake wheel;
secondly, solving a current external parameter matrix of the camera;
and thirdly, calculating a rotation angle.
(1) And extracting and fitting the edge profile of the brake wheel.
The industrial camera is fixed so that the brake wheel is in the center of the camera's field of view. Each frame of image acquired during braking is then processed using a computer. Let a certain frame image be I. Denoising was performed using a 3 × 3 gaussian filter. Then, using the gradient maps f (u, v) obtained by the equations (1) and (2), the threshold value can be set to 25<f(u,v)<255, an edge binary map B as shown in FIG. 4B is obtainedgrad. Color threshold setting to 96 in HLS color space<h<104，170<s<255，nminAnd nmaxSet at 30 and 120 respectively. After color space threshold filtering, a binary image B as shown in FIG. 4c is obtained1. When gradient direction filtering is performed, k is set to 2, and calculation is performed using equation (3), so that a binary map B shown in fig. 4d is obtained2。
Rejecting binary image B2Get a binary map B of the useless detailsrm. The edge is then restored using equation (4) and final 15 is taken, i.e. iterated to X15The result shown in fig. 4e can be obtained. Finally, Canny edge detection is used, and two parameters of a Canny algorithm are 25 and 100 to obtain a single-pixel edge binary image BcannyA single-pixel edge binary image BcannyAnd X15Obtaining a refined edge binary image B by intersectionedgeAs shown in fig. 4 f.
The contour fitting was performed using the least squares method and plotted on a coordinate system, as shown in fig. 5a and 5 b.
(2) Solving a current external parameter matrix of the camera;
firstly, calibrating a camera, firstly placing a calibration plate at a normal calibration position to change the posture for calibration to obtain an internal reference matrix M, and using a formula (15) as a loss function to perform gradient descent. With an rmsprop optimizer, the parameters of rmsprop are: learning rate 0.01, attenuation rate 0.8. And iterating until the value of the loss function is lower than 10 to obtain the calculated external parameter matrix T'.
(3) And (5) calculating the rotation angle. According to the detailed description of the scheme, the brake wheel rotation angle detection process can be determined as follows:
1) the brake sends out a braking and stopping signal and starts to detect;
2) detecting the marker of the kth frame to obtain the coordinate (u) of all the marker images under the coordinate system u-vi,vi) A set of (a);
3) obtaining a world coordinate system x according to the internal and external parameter matrixesw-yw-zwCoordinates of lower (x)i,yi) A set of (a);
4) for all world coordinates, a set S of included angles is obtained using equation (17)k＝{θk,i:i＝1,2,…}；
5) Repeating the steps 1) and 4) for the k +1 th frame to obtain an included angle set Sk+1＝{θk+1,i:i＝1,2,…}；
6) From set Ak,k+1＝{|θk,i-θk+1,j|:θk,i∈Sk,θk+1,j∈Sk+1Find all the smaller than
Figure BDA0002717023210000151
Is averaged to obtain Δk,k+1；
7) The processes from the step 1) to the step 6) are carried out on all the adjacent two frames, and finally, the delta is carried outk,k+1Are summed to obtain
Figure BDA0002717023210000152
Namely the brake wheel corner.

Claims (6)

1. A method for detecting the rotation angle of a brake wheel in the braking process of an elevator brake is characterized by comprising the following steps:
the method comprises the following steps that a plurality of equidistant marks are arranged at positions close to the edges on a brake wheel, an industrial camera is used for shooting the braking process of the brake wheel in the elevator brake while the elevator brake brakes, the shot video image data consisting of n frames of images are transmitted into a computer, and the computer calculates and displays the corner of the brake wheel by using an image processing algorithm, wherein the image processing algorithm comprises the following steps:
the method comprises the following steps of 1, extracting coordinates of edge points of the outer contour of the brake wheel of any frame of image in a video image, and fitting by a least square method to obtain a parameter equation which is met by the contour of the brake wheel in an image coordinate system u-v;
step 2, reversely solving an external parameter matrix T' of the industrial camera according to the parameter equation obtained in the step 1;
step 3, identifying the marker of each frame of image in the video image to obtain the corner of two adjacent frames of images, and setting the corner of the kth frame of image and the (k + 1) th frame of image as deltak,k+1Then angle of rotation Δk,k+1Comprises the following steps:
step 301, detecting the markers of the kth frame image to obtain a set of coordinates of all the markers in an image coordinate system u-v, wherein the coordinates of the ith marker in the kth frame image in the image coordinate system u-v are (u-v)i,vi)；
Step 302, utilizing an inverse matrix T 'of the extrinsic matrix T'-1Transforming the coordinates of all the markers in the k frame image under the image coordinate system u-v into a world coordinate system xw-yw-zwCoordinates of the ith marker in the kth frame image in the world coordinate system xw-yw-zwThe coordinates of (x) belowi,yi)；
Step 303, utilizing the world coordinate system x of each marker in the k frame imagew-yw-zwCalculating the coordinate to obtain the included angle of each marker, and setting the included angle of the ith marker in the kth frame image as thetak,iThen, a set S of all included angles of the k frame image is obtainedk＝{θk,i:i＝1,2,…}；
Step 304, adopting the same steps 301 to 303The method obtains a set S of all included angles of the k +1 th frame imagek+1＝{θk+1,i:i＝1,2,…}；
Step 305, utilizing set Sk＝{θk,iI 1,2, … and set Sk+1＝{θk+1,iI-1, 2, … construct a new set Ak,k+1，Ak,k+1＝{|θk,i-θk+1,j|:θk,i∈Sk,θk+1,j∈Sk+1}；
Step 306, from set Ak,k+1To find out all of
Figure FDA0002717023200000011
The values of (1) are averaged to obtain a rotation angle delta from the k frame image to the (k + 1) th frame imagek,k+1；
And 4, summing the corners of all the two adjacent frames of images to obtain the corner of the brake wheel.
2. The method for detecting the braking wheel angle during the braking process of the elevator brake as claimed in claim 1, wherein said step 1 comprises the steps of:
step 101, after obtaining a gradient map f (u, v) of any frame image I in a video image, taking a threshold value T for the gradient map f (u, v)min<f(u,v)<TmaxTo obtain a binary image Bgrad；
Step 102, for the binary image BgradScreening the points to eliminate points not belonging to the edge of the brake wheel to obtain a binary image B with comprehensively filtered color and gradient thresholds1；
Step 103, binary image B1Contains two edge lines: the gradient direction ranges on two edge lines are different, and the required outer edge is reserved through gradient direction filtering to obtain a binary image B after the gradient direction filtering2；
Step 104, removing the binary image B2Obtaining a binary image B from non-edge detailsrm；
Step 105, the binary image B is processed by the following formularmPerforming edge restoration to obtain a binary image Xfinal：
Figure FDA0002717023200000021
In the formula, X0Representing a binary image obtained by the 0 th iteration; xiRepresenting a binary image obtained by the ith iteration; s is a convolution kernel of image expansion; iterate to XfinalIn the process, the edge can be well restored without introducing unnecessary non-edge details again, and the final value obtains the best effect through actual test;
step 106, for the binary image XfinalEdge thinning is carried out to obtain a single-pixel edge binary image BcannyBy plotting a binary image XfinalAnd a single-pixel edge binary image BcannyObtaining a refined edge binary image B by intersectionedge；
Step 107, obtaining a through-edge binary image B by least square fittingedgeThe obtained contour of the brake wheel satisfies a parametric equation in the image coordinate system u-v.
3. The method for detecting the braking wheel angle during the braking process of the elevator brake as claimed in claim 2, wherein in step 102, the binary image B is processedgradWhen the points in the image are screened, a binary image B is setgradThe coordinate of a certain point A is (u)0,v0) In the image I (u, v) by (u)0,v0) In the region of n × n pixels at the center, if the color value in HLS color space is satisfiedmin<h<hmaxAnd smin<s<smaxThe number of points in the range being greater than nminAnd is less than nmaxThe point A is retained, otherwise from the binary image BgradPoint a is removed.
4. The method for detecting the braking wheel angle during the braking process of an elevator brake as claimed in claim 2, wherein in step 103, the binary image B is taken in an image coordinate system u-v1Middle ladderObtaining the binary image B by the point with the angle of the direction of the degree being in the range of (-arctank, arctank)2And k is the slope corresponding to a half angle of the included angle of the gradient direction range, namely, the following conditions are satisfied:
Figure FDA0002717023200000031
in the formula, graduIs a binary image B1The gradient of any point in the u direction; gradvIs a binary image B1The gradient of any point in the v direction.
5. The method for detecting the rotation angle of the braking wheel of the braking process of the elevator brake as claimed in claim 1, wherein the optical axis of the camera forms an angle α with the rotating shaft of the braking wheel in a transverse plane, and the optical axis of the camera forms an angle β with the rotating shaft of the braking wheel in a longitudinal plane, wherein α is greater than or equal to-20 ° and less than or equal to-20 ° and β is less than or equal to-20 °.
6. The method for detecting the brake wheel angle during the braking process of the elevator brake as claimed in claim 1, wherein the spatial distance from the center of the lens of the camera to the center of the front end face of the elevator brake is d, and d is greater than or equal to 800mm and less than or equal to 1200 mm.
CN202011076670.3A 2020-10-10 2020-10-10 Method for detecting brake wheel corner in braking process of elevator brake Pending CN112027842A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011076670.3A CN112027842A (en) 2020-10-10 2020-10-10 Method for detecting brake wheel corner in braking process of elevator brake

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011076670.3A CN112027842A (en) 2020-10-10 2020-10-10 Method for detecting brake wheel corner in braking process of elevator brake

Publications (1)

Publication Number Publication Date
CN112027842A true CN112027842A (en) 2020-12-04

Family

ID=73573288

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011076670.3A Pending CN112027842A (en) 2020-10-10 2020-10-10 Method for detecting brake wheel corner in braking process of elevator brake

Country Status (1)

Country Link
CN (1) CN112027842A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116703924A (en) * 2023-08-08 2023-09-05 通用电梯股份有限公司 Real-time detection and early warning method for wear state of high-speed elevator parts

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116703924A (en) * 2023-08-08 2023-09-05 通用电梯股份有限公司 Real-time detection and early warning method for wear state of high-speed elevator parts
CN116703924B (en) * 2023-08-08 2023-10-20 通用电梯股份有限公司 Real-time detection and early warning method for wear state of high-speed elevator parts

Similar Documents

Publication Publication Date Title
CN109870461B (en) Electronic components quality detection system
CN103984961B (en) A kind of image detecting method for being used to detect underbody foreign matter
CN106056079B (en) A kind of occlusion detection method of image capture device and human face five-sense-organ
CN112215810B (en) Fatigue test crack monitoring method and device
US20030099330A1 (en) Method for automatically detecting casting defects in a test piece
CN213294338U (en) Brake wheel corner detection device for brake process of elevator brake
CN109297978B (en) Binocular imaging-based power line unmanned aerial vehicle inspection and defect intelligent diagnosis system
CN112027842A (en) Method for detecting brake wheel corner in braking process of elevator brake
CN109117855A (en) Abnormal power equipment image identification system
CN112927223A (en) Glass curtain wall detection method based on infrared thermal imager
CN112634269A (en) Rail vehicle body detection method
CN114708532A (en) Monitoring video quality evaluation method, system and storage medium
CN113553992A (en) Escalator-oriented complex scene target tracking method and system
CN112365516B (en) Virtual and real occlusion processing method in augmented reality
CN109146863A (en) A kind of pavement marker line defect detection device
CN116739998A (en) Coal mine skip roller cage shoe abrasion detection method based on machine vision
KR102355997B1 (en) method for monitoring the soundness of concrete structure based on smart glass
CN114893669A (en) Monitoring device and monitoring method applied to installation of tower crane attachment device
CN115423764A (en) Method and device for detecting shielding of camera and computer equipment
CN112991324A (en) Detection method for visual transmission in magnetic particle detection process
CN112489065A (en) Checkerboard calibration fixed point sub-pixel extraction method
CN111017728A (en) Crane brake health state assessment method based on machine vision
CN114332738B (en) Safety helmet detection system for intelligent construction site
CN110569682A (en) Three-dimensional positioning method for tail end of gantry crane
CN116007953A (en) Abnormality detection method and device for sintering machine trolley

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination