A forklift safety monitoring method and system

Abstract

The present invention relates to the field of forklift safety technology, and solves the technical problem in the prior art that a forklift chain structure is partially deformed and it is difficult to accurately identify the chain deformation only by naked eye recognition, and in particular to a forklift safety monitoring method and system. The method comprises: S1, acquiring an original image of a chain to be detected, and preprocessing the original image to obtain a detection image, S2, constructing a standard curve S according to a standard image containing a chain in a normal form. The present invention uses image processing technology to acquire an image of the front side of the chain, and provides a preliminary rough detection and subsequent accurate defect position detection for chain deformation in different situations, so that the forklift chain can be quickly divided into three categories, thereby improving the work efficiency of detection personnel, solving the problem that it is difficult to accurately identify slight deformation of the chain only by naked eye recognition, and avoiding the cost problem of using expensive visual sensors.

Description

Technical Field

[0001] The present invention relates to the technical field of forklift safety, and in particular to a forklift safety monitoring method and system. Background Art

[0002] A forklift refers to a variety of wheeled transport vehicles that are used for loading and unloading, stacking and short-distance transportation of palletized goods. It is often used for the transportation of large objects in warehouses and is usually driven by a fuel engine or a battery. In the process of using a forklift, several types of problems are often encountered, including power problems, driving problems, steering problems, hydraulic problems and lifting transmission problems. Among them, due to the large size of the fork, the inner transmission parts are blocked. During the work process, the staff tend to neglect the inspection of the inner transmission parts or check them carelessly. Once there is a problem with the fork transmission parts, it is difficult for the inspectors to detect it at the first time. This is also a disadvantage of manual inspection. The existing technology generally uses two methods to perform machine vision monitoring on forklifts. One is to arrange image acquisition equipment around the working area to monitor the forklifts in the working area. The other is to install image acquisition equipment on the forklift to directly perform machine vision monitoring on individual forklifts. It is generally installed on the top of the forklift mast or the top of the cab.

[0003] In the prior art, parts can be inspected manually or through multi-camera visual monitoring methods. For example, in the inspection of the lifting chain connected to the hydraulic system in the fork transmission structure, in actual use, the chain structure may be partially deformed due to heavy overload and bumps and vibrations during movement. It is difficult to accurately identify the slight deformation of the chain only by relying on naked eye recognition, which can easily cause the forklift to stop working due to fork jamming or lifting system damage during operation. Therefore, there is a serious safety hazard. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a forklift safety monitoring method and system, which solves the technical problem that the forks become stuck or the lifting system is damaged and stops working due to the deformation of the lifting chain during the forklift safety monitoring process, and achieves preliminary rough detection and subsequent defect position detection for chain deformation in different situations. The chains can be secondary classified, and the deformed chains due to long-term use can be found in time and the specific location of the chain deformation can be given. At the same time, the chains that need regular maintenance can be given, thereby improving the work efficiency of the inspection personnel and reducing the failure rate caused by the chain.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a forklift safety monitoring method, the method comprising the following steps:

[0006] S1, obtaining the original image of the chain to be detected, and preprocessing the original image to obtain a detection image;

[0007] S2, constructing a standard curve S based on a standard image containing chains of normal morphology;

[0008] S3, extracting a number of detection points from the detection image, and constructing a fitting curve D according to the detection points;

[0009] S4, determining the starting point μ according to the curvature of the standard curve S and the fitting curve D, and establishing a two-dimensional coordinate system with the starting point μ as the origin;

[0010] S5. According to the slope k of any point L on the standard curve S i , and the slope l of point N on the fitting curve D with the same horizontal coordinate as point L i Determine whether the standard curve S and the fitting curve D coincide;

[0011] If k i = l i , then it ends;

[0012] If k i ≠l i , it is marked as a deformed state and enters S6;

[0013] S6. Obtain a combined curve consisting of the standard curve S and the fitting curve D according to the starting point μ, and set the horizontal coordinate range according to the non-overlapping position of the combined curve;

[0014] S7, calculating the error value W between the fitting curve D and the standard curve S within the horizontal coordinate range, and judging whether the fitting curve D is excessively curved according to the error value W;

[0015] If so, mark it as a defective location;

[0016] If not, mark the position as checked and end;

[0017] The step S3 comprises the following steps:

[0018] S31, using the Canny edge detection algorithm to obtain the contour to be detected from the detection image;

[0019] S32, extract all pixel points of the contour to be detected as detection points to obtain a detection point set A(x e ,y f ), and calculate the average value of the vertical coordinates of the detection points The calculation formula is as follows:

[0020]

[0021] Among them, x e Indicates the horizontal coordinate of the e-th detection point, y f represents the ordinate of the fth detection point, and n represents the ordinate y of the detection point fThe quantity, y f represents the ordinate of the f-th detection point;

[0022] S33. Fit multiple detection points to obtain multiple detection curves F c , and calculate the fitting deviation value R used to characterize the goodness of fit 2 ;

[0023] S34. According to the fitting deviation value R 2 obtain the fitting curve D from multiple detection curves F c ;

[0024] If R 2 < E, obtain the fitting curve D and proceed to step S4;

[0025] If R 2 ≥ E, return to step S33 until a fitting curve D that meets the conditions is found;

[0026] The said step S33 includes the following steps:

[0027] S331. Arbitrarily select several detection points A from the detection point set A(x e , y f ), and construct the expression of the fitting curve D according to the contour to be detected. The expression of the fitting curve D is as follows: k D

[0028] D h = ax 2 + bx + c

[0029] where a, b, and c are all curve coefficients, and A k represents the k-th detection point;

[0030] S332. Calculate the values of the coefficients a, b, and c and obtain the fitting curve D. The calculation formula is as follows:

[0031]

[0032] S333. Calculate the fitting deviation value R k according to the fitting value D 2 on the fitting curve D. The calculation formula is as follows:

[0033]

[0034] where D k represents the fitting value on the fitting curve D, y f represents the ordinate value of the detection point, and n represents the logarithm of the selection of the fitting value D k and the ordinate value y f ; Indicates the average value of the vertical coordinate of the detection point, and m indicates the average value With the vertical coordinate value y f The number of pairs selected.

[0035] Preferably, in step S1, the specific implementation steps are as follows:

[0036] S11, using an adaptive median filtering method to perform denoising on the acquired original image to obtain a denoised image;

[0037] S12, segmenting the denoised image using the maximum inter-class variance method to obtain a segmented image;

[0038] S13, using a sub-pixel edge detection algorithm to perform edge detection on the segmented image to obtain a detection image.

[0039] Preferably, in step S2, the specific implementation steps are as follows:

[0040] S21, using a boundary tracing method to obtain a contour image of a qualified chain in a standard image;

[0041] S22, extract qualified pixel point U from the contour image b , and fit the qualified pixels to get the preliminary curve;

[0042] S23, extracting pixel points from the preliminary curve in sequence, and calculating the pixel points on the preliminary curve The qualified pixel point U on the same width b The mean square error value E is calculated as follows:

[0043]

[0044] S24, according to the qualified pixel points, the contour threshold Hy is obtained by using the maximum likelihood estimation method s , determine whether the preliminary curve is the standard curve S;

[0045] If E <Hy s , then the preliminary curve is the standard curve S, and enters S3;

[0046] If E≥Hy s , it is not the standard curve S, and returns to step S21.

[0047] Preferably, in step S4, the specific implementation steps are as follows:

[0048] S41. Determine a pixel point of the standard curve S in the qualified image as a reference point ω, and calculate the curvature value k of the reference point. The calculation formula is as follows:

[0049]

[0050] Where y′ represents the slope of the tangent line at the reference point in the standard curve S, and y ″ represents the second derivative of the reference point in the standard curve S;

[0051] S42, constructing a two-dimensional coordinate system, and mapping the pixel points of the standard curve S in the qualified image to the two-dimensional coordinate system according to the standard size;

[0052] S43, select a point on the fitting curve D with the same curvature value k as a reference point

[0053] S44, mapping the fitting curve D into a two-dimensional coordinate system, and merging the reference point ω and the reference point φ into a starting point μ.

[0054] Preferably, in step S5, the specific implementation steps are as follows:

[0055] S51, select any point L(x i ,y i ), and calculate the slope k of the tangent line at that point i , the calculation formula is as follows:

[0056]

[0057] Among them, dy represents the variable under the differential of the ordinate of point L, and dx represents the variable under the differential of the abscissa of point L;

[0058] S52. Select an x ​​on the fitting curve D that is consistent with point L. i Points N(x i , g i ), and calculate the slope of the tangent line at that point l i , the calculation formula is as follows:

[0059]

[0060] Among them, dg represents the variable under the differential of the horizontal coordinate of point N, x i Indicates the horizontal coordinate value of the i-th point on the standard curve S, y i Indicates the ordinate value of the i-th point on the standard curve S, g i Represents the ordinate value of the i-th point on the fitting curve D;

[0061] S53, judging whether the standard curve S and the fitting curve D overlap;

[0062] If k i = l i , then the standard curve S and the fitting curve D coincide, and the process ends;

[0063] If k i ≠l i, it is marked as the deformed state and proceeds to S6.

[0064] Preferably, in step S6, the abscissa range set according to the non - coincident position of the combined curve is [x r , x t .

[0065] Preferably, in step S7, the specific implementation steps are as follows:

[0066] S71. Calculate the error value W of the ordinates corresponding to each same abscissa on the fitting curve D and the standard curve S within the abscissa range [x r , x t . The calculation formula is as follows:

[0067]

[0068] where z represents the number of abscissas within the range [x r , x t , s h represents the value of the ordinate on the standard curve S, h represents the h - th coordinate, and D h represents the value of the ordinate on the fitting curve D;

[0069] S72. Obtain the deformation threshold B through the maximum likelihood estimation method based on the standard curve S. Determine whether the fitting curve D is overly bent according to the error value W. When the bending degree of the fitting curve D exceeds the limit of the standard curve S, it is overly bent;

[0070] If W≥B, it is marked as the defective position;

[0071] If W < B, it is marked as the inspection position and the process ends;

[0072] S73. Repeat steps S71 - S72 until the fitting curve D is completely detected and all defective positions are obtained.

[0073] This technical solution also provides a system for the above - mentioned forklift safety monitoring method. The system includes: an image acquisition module, which acquires the original image of the chain to be detected and pre - processes the original image to obtain a detection image;

[0074] A first image processing module, which is used to construct the standard curve S according to the standard image containing the chain in the normal state;

[0075] A second image processing module, which is used to extract several detection points from the detection image and construct the fitting curve D according to the detection points;

[0076] An image fitting module, which is used to determine the starting point μ according to the curvatures of the standard curve S and the fitting curve D, And and establish a two - dimensional coordinate system with the starting point μ as the origin;

[0077] The first analysis module is used to determine the slope k of any point L on the standard curve S. i , and the slope l of point N on the fitting curve D with the same horizontal coordinate as point L i Determine whether the standard curve S and the fitting curve D coincide;

[0078] The second analysis module is used to obtain a combined curve consisting of a standard curve S and a fitting curve D according to the starting point μ, and to set a horizontal coordinate range according to a non-overlapping position of the combined curve;

[0079] The third analysis module is used to calculate the error value W between the fitting curve D and the standard curve S within the horizontal coordinate range, and determine whether the fitting curve D is excessively curved according to the error value W.

[0080] Through the above technical solution, the present invention provides a forklift safety monitoring method and system, which has at least the following beneficial effects:

[0081] 1. The present invention uses image processing technology to acquire images of the front side of the chain, and provides preliminary rough detection and subsequent accurate defect position detection for chain deformation in different situations, so that the forklift chain can be quickly divided into three categories, thereby improving the work efficiency of the inspectors. It solves the problem in the prior art that it is difficult to accurately identify the slight deformation of the chain by relying solely on naked eye recognition, which easily causes the forklift to stop working due to fork jamming or lifting system damage during operation, and also avoids the cost problem of using equipment such as multi-camera visual sensors.

[0082] 2. The present invention performs error detection based on the non-overlapping positions of the standard curve and the fitting curve in the two-dimensional coordinate system, and uses a combination of mathematical methods and visual processing to calculate the error value and give the defect position and the regular inspection position, so that the inspection personnel can directly arrange the work according to the inspection results, thereby improving the work efficiency of the inspection personnel and making the inspection results more detailed and targeted. The specific position and number of the defect position can be obtained, and the position that needs regular inspection will not be ignored, thereby avoiding safety accidents caused by negligence in manual inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0084] Figure 1 is a flow chart of the security monitoring method of the present invention;

[0085] Figure 2 It is a comparison diagram of the standard image and the image with detection of the present invention;

[0086] Figure 3 It is a schematic diagram of a standard curve in the standard image of the present invention;

[0087] Figure 4 It is a schematic diagram of a fitting curve in a detection image of the present invention;

[0088] Figure 5 It is a schematic diagram of the standard curve and the fitting curve of the present invention in the coordinate system;

[0089] Figure 6 Schematic diagram of a forklift safety monitoring system of the present invention. DETAILED DESCRIPTION

[0090] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific implementation methods, so that the implementation process of how the present application uses technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.

[0091] In this embodiment, since the prior art generally performs image monitoring on the forklift working area, or is installed on the forklift to monitor the driving and driver operation of the forklift, there is a lack of monitoring of the forklift parts. However, in the process of long-term use of the forklift for operation, due to the extremely heavy weight of the cargo, the bumps and vibrations caused by the increase and decrease of the forklift's driving acceleration, and the long-term use every day, the forklift parts are very likely to age and deform, which can easily cause the forklift to stop working or be forced to repair, seriously affecting normal operation, especially the chain connected to the fork and the hydraulic system, which is more prone to deformation. Once it is not maintained or replaced in time, the lifting of the fork cannot be carried out normally, the entire forklift cannot be put into work, and there will be safety hazards. In order to solve this problem, the present invention uses image processing technology to acquire images from the front of the chain, and provides preliminary rough detection and subsequent defect position detection for chain deformation in different situations, so that the chain can be secondary classified, and the chain deformed due to long-term use can be found in time, and the specific location of the chain deformation can be given, and the chain that needs regular maintenance can be given.

[0092] Please refer to Figure 1-6 This embodiment provides a forklift safety monitoring method to solve the problem that the lifting chain structure of the forklift cannot be monitored in time and the position of structural deformation cannot be identified. The method includes the following steps:

[0093] S1. Obtain the original image of the chain to be detected, and pre-process the original image to obtain the detection image; in this step, since the original image is inevitably interfered by various random noises during the acquisition process, it is necessary to denoise the original image. Since the original image cannot be directly processed over a large area, it is necessary to use the distribution of the original image pixels to process the original image, segment the required part, and finally use the pixels with large gray value changes to detect the edges in the image, which is convenient for later image recognition. In order to facilitate the implementation of this step, the specific steps are as follows:

[0094] S11, using an adaptive median filtering method to perform denoising on the acquired original image to obtain a denoised image;

[0095] S12, segmenting the denoised image using the maximum inter-class variance method to obtain a segmented image;

[0096] S13, performing edge detection on the segmented image using a sub-pixel edge detection algorithm to obtain a detection image;

[0097] It should be noted in step S1 that, in actual operation, it is found that after using mean filtering and Gaussian filtering to denoise the image, although the operation is simple, efficient and can effectively reduce noise, when the filter template continues to increase, it is easy to cause image blur. The median filtering algorithm will bring the same problem. In image segmentation, the maximum entropy threshold segmentation algorithm is sensitive to noise, which results in unsatisfactory image segmentation effect. Compared with the conventional pixel edge detection algorithm, the sub-pixel edge detection algorithm is more accurate. The corresponding results can be obtained by calculating using the existing adaptive median filtering method, the maximum inter-class variance method and the sub-pixel edge detection algorithm. No improvement is required, so it will not be described in detail.

[0098] S2. Construct a standard curve S based on a standard image containing a normal morphological chain. To facilitate the implementation of this step, the specific steps are as follows:

[0099] S21. A contour image of a qualified chain in a standard image is obtained by using a boundary tracking method. In this step, it should be noted that the boundary tracking method is a method of tracking boundary points in an image in a certain direction using pixel values ​​of pixels in an image and taking one of the fixed points as a starting point. The boundary tracking method is commonly used in image processing and has the advantages of strong adaptability, good real-time performance, easy implementation and expansion. The present invention uses the existing boundary tracking method to calculate and obtain the corresponding result without any improvement, so it will not be described in detail.

[0100] S22, extract qualified pixel point U from the contour image b , and fit the qualified pixels to get the preliminary curve;

[0101] S23, extracting pixel points from the preliminary curve in sequence, and calculating the pixel points on the preliminary curve The qualified pixel point U on the same width b The mean square error value E is calculated as follows:

[0102]

[0103] S24, according to the qualified pixel points, the contour threshold Hy is obtained by using the maximum likelihood estimation method s , determine whether the preliminary curve is the standard curve S;

[0104] If E <Hy s , then the preliminary curve is the standard curve S, and enters S3;

[0105] If E≥Hy s , it is not the standard curve S, and returns to step S21. In step S24, it should be noted that the maximum likelihood estimation method is a method for estimating the function threshold using statistics, which is more common in practice. Its principle is to write the likelihood function according to the curve expression, solve the likelihood function by the derivation method and obtain the likelihood function estimate, and use the function estimate to obtain the deformation threshold. The present invention uses the existing maximum likelihood estimation method to calculate and obtain the corresponding result, and no improvement is required, so it will not be described in detail.

[0106] In actual work, chains generally experience compression deformation and tensile deformation during long-term use. Compression deformation and tensile deformation are caused by the chain structure being subjected to a large impact during the bumpy process of the forklift, which causes the chain to twist. If no maintenance measures are taken, not only will it not recover during use, but it will easily cause the fork to jam during lifting. To solve the above problems, manual identification is required by staff, and the low efficiency of manual identification and the accuracy of naked eye identification bring great problems to the detection. The present invention utilizes the fact that the deformed chain will undergo slight deformation, and obvious contour changes are displayed in the image, forming contours with different slopes at a certain point on the arc curve, and further discovers the difference between the fitting curve D and the standard curve S, and then distinguishes the image for the first time, marks the fitting curve D that does not coincide with the standard curve S, and also facilitates the second distinction later. In order to facilitate the implementation of this step, the specific steps are as follows:

[0107] S3. Extract several detection points from the detection image and construct a fitting curve D according to the detection points. To facilitate the implementation of this step, the specific implementation steps are as follows:

[0108] S31, using the Canny edge detection algorithm to obtain the contour to be detected from the detection image;

[0109] S32, extract all pixel points of the contour to be detected as detection points to obtain a detection point set A(x e ,y f ), and calculate the average value of the vertical coordinates of the detection points The calculation formula is as follows:

[0110]

[0111] Among them, x e Indicates the horizontal coordinate of the e-th detection point, y f represents the ordinate of the fth detection point, and n represents the ordinate y of the detection point f The number of f represents the ordinate of the fth detection point;

[0112] S33, fitting any number of detection points to obtain multiple detection curves F c , and calculate the fitting partial value R used to characterize the goodness of fit 2 In this step, at least 21 pixels are obtained on the contour to be detected, and at least 3 detection curves F are obtained. c In actual operation, due to the limitation of computing power, it is fastest and more effective to obtain 8 curves and make judgments. In order to achieve this step, the specific implementation steps are as follows:

[0113] S331, from the detection point set A(x e ,y f ) randomly select several detection points A k , construct the expression of fitting curve D according to the contour to be detected, and the expression of fitting curve D is as follows:

[0114] D h =ax 2 +bx+c

[0115] Among them, a, b, and c are curve coefficients, A k represents the kth detection point;

[0116] S332, calculate the values ​​of coefficients a, b, c and obtain the fitting curve D, the calculation formula is as follows:

[0117]

[0118] S333, according to the fitting value D on the fitting curve D k Calculate the fitted partial value R 2 , the calculation formula is as follows:

[0119]

[0120] Among them, D k represents the fitting value on the fitting curve D, y frepresents the ordinate value of the detection point, and n represents the fitting value D k and the ordinate value y f The logarithm selected represents the average value of the ordinate of the detection point, and m represents the average value and the ordinate value y f The logarithm selected. It should be noted in this step that the expression of the fitting curve D is only an example of an arc line formed by a quadratic equation. In actual operation, for different chain arc surfaces, the polynomial fitting method can be applied to obtain the polynomial expression by testing the order of the polynomial reality.

[0121] S34. According to the fitting deviation value R 2 Obtain the fitting curve D from multiple detection curves F c ;

[0122] If R 2 < E, the fitting curve D is obtained and step S4 is entered;

[0123] If R 2 ≥ E, return to step S33 until a fitting curve D that meets the conditions is found.

[0124] S4. Determine the starting point μ according to the curvatures of the standard curve S and the fitting curve D, and establish a two-dimensional coordinate system with the starting point μ as the origin; in this step, use the points with equal curvatures as the starting point μ, and use the starting point μ to establish the two-dimensional coordinate systems of the standard curve S and the fitting curve D respectively. The specific implementation steps are as follows:

[0125] S41. Determine a pixel point of the standard curve S in the qualified image as the reference point ω, and calculate the curvature value k of the reference point. The calculation formula is as follows:

[0126]

[0127] where y′ represents the tangent slope of the reference point in the standard curve S, and y″ represents the second derivative of the reference point in the standard curve S;

[0128] S42. Construct a two-dimensional coordinate system, and map the pixel points of the standard curve S in the qualified image to the two-dimensional coordinate system according to the standard size;

[0129] S43. Select a point on the fitting curve D with the same curvature value k as the reference point

[0130] S44. Map the fitting curve D to the two-dimensional coordinate system, and make the reference point ω and the reference point φ coincide to become the starting point μ.

[0131] S5. According to the slope k of any point L on the standard curve S i, and the slope l of point N on the fitting curve D with the same horizontal coordinate as point L i Determine whether the standard curve S and the fitting curve D coincide;

[0132] If k i = l i , then it ends;

[0133] If k i ≠l i , it is marked as a deformed state and enters S6;

[0134] In this step, the slope of the points on the same worksheet on the standard curve S and the fitting curve D is compared by using the change characteristics of the slope of the points on the curve, and then it is determined whether the standard curve S and the fitting curve D overlap. The specific implementation steps are as follows:

[0135] S51, select any point L(x i ,y i ), and calculate the slope k of the tangent line at that point i , the calculation formula is as follows:

[0136]

[0137] Among them, dy represents the variable under the differential of the ordinate of point L, and dx represents the variable under the differential of the abscissa of point L;

[0138] S52. Select an x ​​on the fitting curve D that is consistent with point L. i Points N(x i , g i ), and calculate the slope of the tangent line at that point l i , the calculation formula is as follows:

[0139]

[0140] Among them, dg represents the variable under the differential of the horizontal coordinate of point N, x i Indicates the horizontal coordinate value of the i-th point on the standard curve S, y i Indicates the ordinate value of the i-th point on the standard curve S, g i Represents the ordinate value of the i-th point on the fitting curve D;

[0141] S53, judging whether the standard curve S and the fitting curve D overlap;

[0142] If k i = l i , then the standard curve S and the fitting curve D coincide, and the process ends;

[0143] If k i ≠l i, it is marked as a deformed state and enters S6.

[0144] It should be noted that in the above steps, the common manual periodic inspection method for inspecting the forklift chain in actual operation can ensure that the forklift avoids problems during use by manually checking the chain in sequence. However, the quality of manual inspection is based on the carefulness of the inspector, and the inspection method requires the inspector to perform inspection when the forklift is stopped, which not only delays the work progress but also wastes a lot of time. Some factories also choose to use visual sensors to monitor the safety of forklifts, but the use cost and maintenance cost of visual sensors are relatively high. The present invention uses an image acquisition method to make a preliminary distinction based on the fitting curve D of the chain edge in the detection image, and classifies normal chains and chains with bending deformation. It is not only fast and low-cost, but also improves work efficiency and facilitates the second inspection and classification in the later stage. In actual operation, in order to make the image more distinguishable, softer diffuse reflected light is used for illumination to make the image itself clearer.

[0145] In the deformation of the chain, according to the different degrees of deformation, it is found that some chains with slight deformation can work normally in actual work and can be used for a period of time, while other chains with large deformation degrees make the forklift unable to operate normally in actual work, and there will be jams or failure to work. The present invention further classifies the deformed chains and detects the defective positions with large deformation degrees, which is convenient for the inspection personnel to replace and improves the work efficiency of the inspection personnel. In order to realize this function, the specific implementation steps are as follows:

[0146] S6. Obtain a combined curve consisting of the standard curve S and the fitting curve D according to the starting point μ, and set the horizontal coordinate range according to the non-overlapping position of the combined curve, wherein the horizontal coordinate range set at the non-overlapping position is [x r , x t ];

[0147] S7, calculating the error value W between the fitting curve D and the standard curve S within the horizontal coordinate range, and judging whether the fitting curve D is excessively curved according to the error value W;

[0148] If so, mark it as a defective location;

[0149] If not, mark the position as checked and end.

[0150] In this step, the difference between the fitting curve D and the standard curve S is determined by the error value W between the fitting curve D and the standard curve S within the horizontal axis range, and then the degree of deformation of the fitting curve D can be known. The position and number of defects can be finally known by marking the deformation position. The specific implementation steps are as follows:

[0151] S71. Calculate the error value W of the ordinates corresponding to each same abscissa on the fitting curve D and the standard curve S within the abscissa range [x r , x t . The calculation formula is as follows:

[0152]

[0153] where z represents the number of abscissas within the range [x r , x t , s h represents the value of the ordinate on the standard curve S, h represents the h-th coordinate, and D h represents the value of the ordinate on the fitting curve D;

[0154] S72. Obtain the deformation threshold B through the maximum likelihood estimation method based on the standard curve S. Determine whether the fitting curve D is overly curved according to the error value W. When the degree of curvature of the fitting curve D exceeds the limit of the standard curve S, it is considered overly curved;

[0155] If W≥B, mark it as the defect position;

[0156] If W<B, mark it as the inspection position and end;

[0157] S73. Repeat steps S71 - S72 until the fitting curve D is completely detected and all defect positions are obtained.

[0158] Corresponding to the forklift safety monitoring method provided in the above embodiment, this embodiment also provides a system for the forklift safety monitoring method. Since the forklift safety monitoring system provided in this embodiment corresponds to the forklift safety monitoring method provided in the above embodiment, the implementation manners of the foregoing forklift safety monitoring method are also applicable to the forklift safety monitoring system provided in this embodiment and will not be described in detail in this embodiment.

[0159] Please refer to Figure 6 , which shows the structural block diagram of the forklift safety monitoring system provided in this embodiment. The forklift safety monitoring system includes an image acquisition module, a first image processing module, a second image processing module, an image fitting module, a first analysis module, a second analysis module, and a third analysis module.

[0160] The image acquisition module obtains the original image of the chain to be detected and preprocesses the original image to obtain the detection image. The first image processing module is used to construct a standard curve S according to the standard image containing the normal morphology chain. The second image processing module is used to extract a number of detection points from the detection image and construct a fitting curve D according to the detection points. The image fitting module is used to determine the starting point μ according to the curvature of the standard curve S and the fitting curve D, and establish a two-dimensional coordinate system with the starting point μ as the origin. The first analysis module is used to determine the slope k of any point L on the standard curve S. i , and the slope l of point N on the fitting curve D with the same horizontal coordinate as point L i Determine whether the standard curve S and the fitting curve D overlap. The second analysis module is used to obtain a combined curve composed of the standard curve S and the fitting curve D according to the starting point μ, and set the horizontal coordinate range according to the non-overlapping position of the combined curve. The third analysis module is used to calculate the error value W between the fitting curve D and the standard curve S within the horizontal coordinate range, and determine whether the fitting curve D is excessively curved according to the error value W.

[0161] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, so the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0162] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the above embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0163] The above implementation methods have been described in detail. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A forklift safety monitoring method, characterized in that: The method includes the following steps: S1. Obtain the original image of the chain to be detected, and preprocess the original image to obtain a detection image; S2. Construct a standard curve S according to the standard image including the chain in normal form; S3. Extract several detection points from the detection image, and construct a fitting curve D according to the detection points; S4. Determine the starting point μ according to the curvatures of the standard curve S and the fitting curve D, and establish a two-dimensional coordinate system with the starting point μ as the origin; S5. According to the slope k of any point L on the standard curve S i , and the slope l of point N on the fitting curve D with the same horizontal coordinate as point L i Determine whether the standard curve S and the fitting curve D coincide; If k i = l i , then it ends; If k i ≠l i , it is marked as a deformed state and enters S6; S6. Obtain the combined curve formed by the standard curve S and the fitting curve D according to the starting point μ, and set the abscissa range according to the non-coincident positions of the combined curve; S7. Calculate the error value W between the fitting curve D and the standard curve S within the abscissa range, and judge whether the fitting curve D is excessively bent according to the error value W; If so, mark it as a defective position; If not, mark it as an inspection position and end; The step S3 includes the following steps: S31. Use the Canny edge detection algorithm to obtain the contour to be detected from the detection image; S32, extract all pixel points of the contour to be detected as detection points to obtain a detection point set A(x e ,y f ), and calculate the average value of the vertical coordinates of the detection points The calculation formula is as follows: Among them, x e Indicates the horizontal coordinate of the e-th detection point, y f represents the ordinate of the fth detection point, and n represents the ordinate y of the detection point f The number of f represents the ordinate of the fth detection point; S33, fitting any number of detection points to obtain multiple detection curves F c , and calculate the fitting partial value R used to characterize the goodness of fit 2 ; S34, according to the fitting bias R 2 From multiple detection curves F c The fitting curve D is obtained; If R 2 < is less than E, a fitted curve D is obtained and step S4 is entered; If R 2 ≥E, then return to step S33 until a fitting curve D that meets the conditions is found; The step S33 includes the following steps: S331, from the detection point set A(x e ,y f ) randomly select several detection points A k , construct the expression of the fitting curve D according to the contour to be detected, and the expression of the fitting curve D is as follows: D h =ax 2 +bx+c Among them, a, b, and c are curve coefficients, A k represents the kth detection point; S332. Calculate the numerical values of coefficients a, b, and c and obtain the fitting curve D. The calculation formula is as follows: S333, according to the fitting value D on the fitting curve D k Calculate the fitted partial value R 2 , the calculation formula is as follows: Among them, D k represents the fitting value on the fitting curve D, y f represents the ordinate value of the detection point, and n represents the fitting value D k With the vertical coordinate value y f The number of selected logarithms, Indicates the average value of the vertical coordinate of the detection point, and m indicates the average value With the vertical coordinate value y f The number of pairs selected.

2. The monitoring method according to claim 1, characterized in that: In step S1, the specific implementation steps are as follows: S11. Use the adaptive median filtering method to denoise the obtained original image to obtain a denoised image; S12. Use the Otsu method to segment the denoised image to obtain a segmented image; S13. Use the sub-pixel edge detection algorithm to perform edge detection on the segmented image to obtain a detection image.

3. The monitoring method according to claim 1, characterized in that: In step S2, the specific implementation steps are as follows: S21. Use the boundary tracking method to obtain the contour image of the qualified chain in the standard image; S22, extract qualified pixel point U from the contour image b , and fit the qualified pixels to get the preliminary curve; S23, extracting pixel points from the preliminary curve in sequence, and calculating the pixel points on the preliminary curve The qualified pixel point U on the same width b The mean square error value E is calculated as follows: S24, according to the qualified pixel points, the contour threshold Hy is obtained by using the maximum likelihood estimation method s , determine whether the preliminary curve is the standard curve S; If E <Hy s , then the preliminary curve is the standard curve S, and enters S3; If E≥Hy s , it is not the standard curve S, and returns to step S21.

4. The monitoring method according to claim 1, characterized in that: In step S4, the specific implementation steps are as follows: S41. Determine a pixel point of the standard curve S in the qualified image as the reference point ω, and calculate the curvature value k of the reference point. The calculation formula is as follows: where y′ represents the tangent slope of the reference point in the standard curve S, and y″ represents the second derivative of the reference point in the standard curve S; S42. Construct a two-dimensional coordinate system, and map the pixel points of the standard curve S in the qualified image to the two-dimensional coordinate system according to the standard size; S43, select a point on the fitting curve D with the same curvature value k as a reference point S44. Map the fitting curve D to the two-dimensional coordinate system, and make the reference point ω and the reference point φ coincide to become the starting point μ.

5. The monitoring method according to claim 1, characterized in that: In step S5, the specific implementation steps are as follows: S51, select any point L(x i ,y i ), and calculate the slope k of the tangent line at that point i , the calculation formula is as follows: where dy represents the variable under the differential of the ordinate of point L, and dx represents the variable under the differential of the abscissa of point L; S52. Select an x ​​on the fitting curve D that is consistent with point L. i Points N(x i , g i ), and calculate the slope of the tangent line at that point l i , the calculation formula is as follows: Among them, dg represents the variable under the differential of the horizontal coordinate of point N, x i Indicates the horizontal coordinate value of the i-th point on the standard curve S, y i Indicates the ordinate value of the i-th point on the standard curve S, g i Represents the ordinate value of the i-th point on the fitting curve D; S53. Judge whether the standard curve S and the fitting curve D coincide; If k i = l i , then the standard curve S and the fitting curve D coincide, and the process ends; If k i ≠l i , it is marked as a deformed state and enters S6.

6. The monitoring method according to claim 1, characterized in that: In step S6, the horizontal coordinate range is set as [x r , x t ].

7. The monitoring method according to claim 1, characterized in that: In step S7, the specific implementation steps are as follows: S71, calculate the horizontal coordinate range [x r , x t The error value W of the ordinate corresponding to each same abscissa on the fitting curve D and the standard curve S is calculated as follows: Among them, z represents the range [x r , x t ]The number of horizontal coordinates in s h represents the value of the ordinate on the standard curve S, h represents the hth coordinate, D h Represents the value of the ordinate on the fitting curve D; S72. Obtain the deformation threshold B according to the standard curve S through the maximum likelihood estimation method, and judge whether the fitting curve D is excessively bent according to the error value W. When the bending degree of the fitting curve D exceeds the limit of the standard curve S, it is considered excessively bent; If W≥B, mark it as a defective position; If W<B, mark it as an inspection position and end; S73. Repeat steps S71 - S72 until the fitting curve D is completely detected and all defective positions are obtained.

8. A system for implementing the forklift safety monitoring method according to any one of claims 1 to 7, characterized in that: The system includes: An image acquisition module, which obtains the original image of the chain to be detected, and preprocesses the original image to obtain a detection image; A first image processing module, used for constructing a standard curve S according to a standard image containing a normal morphological chain; A second image processing module is used to extract a number of detection points from the detection image and construct a fitting curve D according to the detection points; The image fitting module is used to determine the starting point μ according to the curvature of the standard curve S and the fitting curve D. And The starting point μ is used as the origin to establish a two-dimensional coordinate system; The first analysis module is used to determine the slope k of any point L on the standard curve S. i , and the slope l of point N on the fitting curve D with the same horizontal coordinate as point L i Determine whether the standard curve S and the fitting curve D coincide; The second analysis module is used to obtain a combined curve consisting of a standard curve S and a fitting curve D according to the starting point μ, and to set a horizontal coordinate range according to a non-overlapping position of the combined curve; The third analysis module is used to calculate the error value W between the fitting curve D and the standard curve S within the horizontal coordinate range, and determine whether the fitting curve D is excessively curved according to the error value W.

Images