Elevator sill surface wear degree identification method based on computer vision

Abstract

The invention relates to the technical field of elevator wear detection, in particular to an elevator sill surface wear degree identification method based on computer vision, and the method comprises the steps: obtaining a surface image of an elevator sill, and carrying out the graying of the surface image, and obtaining a corresponding gray image; performing full threshold segmentation on the grayscale image by using at least two preset grayscale thresholds to correspondingly obtain at least two segmented images; forming a longitudinal pixel point sequence by each column of pixel points in any segmented image, and obtaining the local wear degree corresponding to any segmented image according to the gray value distribution characteristics of each longitudinal pixel point sequence; the local wear degree corresponding to each segmented image is obtained, and the global wear degree of the elevator sill in the gray level image is obtained according to the local wear degrees corresponding to all the segmented images; and the surface wear result of the elevator sill is determined according to the global wear degree, so that the surface wear evaluation of the elevator sill is more accurate and efficient.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator wear detection, and in particular to a method for identifying the wear degree of an elevator sill surface based on computer vision. Background Art

[0002] The elevator sill is a metal pedal with grooves for entering and exiting the elevator car or floor door. It is responsible for the normal sliding of the car door slider and the footsteps of passengers entering and exiting the door. With the popularity of elevators and the increase in daily usage, daily inspection and maintenance of elevators are necessary. However, the wear of the elevator sill will affect the smoothness of the door guide rail, causing the elevator door to not close tightly, affecting the safety of operation. Secondly, the surface wear of the elevator sill may also cause passengers' shoes, wheelchairs, clothes, etc. to be stuck, causing falls or clamping accidents. Therefore, the surface wear degree of the elevator sill is also a necessary inspection item for daily monitoring and maintenance of elevators.

[0003] At present, computer vision-based detection technology has been widely used in target defect detection. The existing technology can collect the surface image of the elevator sill through the camera and determine the possible surface wear area through the threshold segmentation algorithm. However, the top of the elevator is usually equipped with lights. Affected by the lighting environment in the elevator, the overall grayscale distribution in the surface image of the elevator sill taken is uneven. The grayscale of the area directly illuminated by the light is relatively high, and the grayscale of other areas is relatively low. If there is a wear area, the wear area will also have grayscale differences due to the lighting, making it difficult to obtain the wear defect area through threshold segmentation, and thus unable to accurately identify the surface wear defects of the elevator sill.

[0004] Therefore, how to accurately identify the surface wear defects of elevator sills has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method for identifying the degree of surface wear of an elevator sill based on computer vision to solve the problem of how to accurately identify the surface wear defects of an elevator sill.

[0006] An embodiment of the present invention provides a method for identifying the degree of wear on the surface of an elevator sill based on computer vision, the method comprising the following steps: Acquire a surface image of an elevator sill, wherein the direction of the elevator sill in the surface image is parallel to the longitudinal edge of the image, grayscale the surface image to obtain a corresponding grayscale image; perform full threshold segmentation on the grayscale image using at least two preset grayscale thresholds, and obtain at least two segmented images accordingly; For any segmented image, according to the shape characteristics of the elevator sill, each column of pixels in the segmented image is respectively formed into a longitudinal pixel sequence, and according to the gray value distribution characteristics of each longitudinal pixel sequence, the wear degree characteristic value of each longitudinal pixel sequence is correspondingly obtained, and according to the wear degree characteristic values ​​of all longitudinal pixel sequences, the local wear degree corresponding to any segmented image is obtained; The local wear degree corresponding to each of the segmented images is obtained, and the global wear degree of the elevator sill in the grayscale image is obtained according to the local wear degrees corresponding to all the segmented images; and the surface wear result of the elevator sill is determined according to the global wear degree.

[0007] In one embodiment, the full threshold segmentation means that the grayscale value greater than the grayscale threshold is set to 255, and the grayscale value less than or equal to the grayscale threshold is set to 0, then the grayscale value distribution characteristics of each longitudinal pixel point sequence are correspondingly obtained, including: For any longitudinal pixel point sequence, pixels with continuous grayscale values ​​of 0 in any longitudinal pixel point sequence are grouped into pixel point sets to obtain at least two pixel point sets, and the number of interval pixel points between each two adjacent pixel point sets is obtained according to the position of the last pixel point in each of the pixel point sets in any longitudinal pixel point sequence to obtain the total number of interval pixel points and the variance of the number of interval pixel points; The number of pixels in each pixel point set is multiplied to obtain a cumulative value, the cumulative value is normalized to obtain a first normalized value, the product between the total number of interval pixels and the variance of the number of interval pixels is normalized to obtain a second normalized value, and the wear degree characteristic value of any longitudinal pixel point sequence is obtained based on the sum of the first normalized value and the second normalized value.

[0008] In one embodiment, obtaining the local wear degree corresponding to any segmented image according to the wear degree characteristic values ​​of all longitudinal pixel point sequences includes: Obtaining a preset wear degree characteristic value threshold, and taking a longitudinal pixel point sequence whose wear degree characteristic value is greater than or equal to the wear degree characteristic value threshold as a wear characteristic sequence; The wear feature similarity of each wear feature sequence is obtained, the number of pixels with a gray value of 0 in each wear feature sequence is counted, and the local wear degree corresponding to any segmented image is obtained according to the wear feature similarity and the number of pixels of each wear feature sequence.

[0009] In one embodiment, obtaining the wear feature similarity of each wear feature sequence includes: For any wear feature sequence, obtain the left adjacent longitudinal pixel point sequence and the right adjacent longitudinal pixel point sequence of the any wear feature sequence from all longitudinal pixel point sequences, obtain the absolute values ​​of the wear degree characteristic value differences between the any wear feature sequence and the left adjacent longitudinal pixel point sequence and the right adjacent longitudinal pixel point sequence, respectively, accumulate all the absolute values ​​of the wear degree characteristic value differences to obtain the accumulated value, and use the inverse of the accumulated value as the wear feature similarity of the any wear feature sequence.

[0010] In one embodiment, obtaining the local wear degree corresponding to any segmented image according to the wear feature similarity and the number of pixels of each wear feature sequence includes: According to the wear feature similarity of each wear feature sequence, a weight is set for each wear feature sequence. The greater the wear feature similarity, the greater the weight. The number of pixel points of all wear feature sequences is weighted and summed to obtain the local wear degree corresponding to any segmented image.

[0011] In one embodiment, obtaining the global wear degree of the elevator sill in the grayscale image according to the local wear degrees corresponding to all segmented images includes: For any wear feature sequence of any segmented image, the position of any wear feature sequence in any segmented image is taken as the target position, and the number of other segmented images whose longitudinal pixel point sequence at the target position in each other segmented image is the wear feature sequence is counted and recorded as the same number; Obtain the same number corresponding to each wear feature sequence in any of the segmented images, obtain the total number of the same number of any of the segmented images, obtain the total number of the same number of each of the segmented images, accumulate the product of the local wear degree of all segmented images and the total number of the same number, and obtain the global wear degree of the elevator sill in the grayscale image.

[0012] In one embodiment, determining the surface wear result of the elevator sill according to the global wear degree includes: The global wear degree is normalized to obtain a normalized wear value, and when the normalized wear value is greater than or equal to a preset severe wear value threshold, it is determined that the surface of the elevator sill is severely worn.

[0013] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention evaluates the wear condition of the elevator sill through multiple thresholds. Specifically, in the segmented image under each threshold, the segmented image is divided into multiple longitudinal regions (longitudinal pixel point sequences) according to the shape characteristics of the elevator sill, and then the distribution characteristics of each longitudinal region are analyzed separately to obtain a relatively accurate local wear degree under the shadow background caused by illumination. While avoiding the difficulty in segmenting the wear area due to the influence of the elevator light, the grayscale difference between different areas caused by illumination can be eliminated, and all local wear features of the elevator sill surface can be obtained without omission. Then, the global wear degree of the elevator sill is obtained by combining the local wear degree of each segmented image under multiple thresholds, which is used to evaluate the actual wear condition of the elevator sill, making the surface wear evaluation of the elevator sill simpler and more efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0015] Figure 1 is a method flow chart of a method for identifying the degree of wear on the surface of an elevator sill based on computer vision provided in Embodiment 1 of the present invention; Figure 2 is a schematic diagram of a surface image of an elevator sill provided by an embodiment of the present invention; Figure 3 is a schematic diagram of a segmented image of an elevator sill corresponding to different grayscale thresholds provided by an embodiment of the present invention; Figure 4 It is a schematic diagram of the connection between a shadow background and a low grayscale area provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0017] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0018] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0019] See also Figure 1 , is a method flow chart of a method for identifying the degree of wear of an elevator sill surface based on computer vision provided in Embodiment 1 of the present invention, such as Figure 1 As shown, the method may include: Step S101, obtaining a surface image of an elevator sill, wherein the direction of the elevator sill in the surface image is parallel to the longitudinal edge of the image, graying the surface image to obtain a corresponding grayscale image; performing full threshold segmentation on the grayscale image using at least two preset grayscale thresholds, and correspondingly obtaining at least two segmented images.

[0020] Use the camera on the top of the elevator to shoot the elevator sill. Take the image of the elevator sill once every 12 hours to get a surface image of the elevator sill. The collected surface image is referenced Figure 2 As shown, the direction of the elevator sill in the surface image is parallel to the longitudinal edge of the image. Afterwards, the collected surface image is transmitted to the data processing terminal, which is used for the terminal to analyze the surface image and realize the identification of the wear of the elevator sill.

[0021] Due to the influence of the elevator light, different areas of the elevator sill will present different brightness distributions. Secondly, considering that when the existing threshold segmentation algorithm is used to identify the wear area in the surface image of the elevator sill, the influence of light will cause large errors in the wear identification results. Therefore, in order to improve the accuracy of wear identification based on threshold segmentation, in an embodiment of the present invention, a multi-threshold segmentation method is combined to perform wear identification analysis.

[0022] Specifically, the surface image is first grayed to obtain a gray image of the drinking surface, and then the gray image is segmented using a full threshold segmentation method, wherein the full threshold segmentation method means that the gray value greater than the gray threshold is set to 255, and the gray value less than or equal to the gray threshold is set to 0. Figure 3, which is the segmented image of the elevator sill corresponding to different grayscale thresholds, wherein the grayscale thresholds include 30, 50, 70, 90, 110, 120, 130, 150, 160, 170, 180, 190. As the grayscale threshold increases, the low-grayscale pixels in the segmented image gradually appear, and the smaller the threshold difference between different grayscale thresholds, the smaller the gradient between the segmented images, and the more capable of capturing defect details. Therefore, in order to obtain segmented images under multiple grayscale thresholds, in an embodiment of the present invention, a moderate value is set, that is, the threshold difference between the multiple grayscale thresholds is set to 10, which can retain the wear defect details while avoiding excessive calculation. Then, starting from the grayscale value of 0, the threshold difference of 10 is gradually increased to obtain multiple grayscale thresholds in the range of 0-255. There is no restriction here, and it can be set according to the implementation scenario.

[0023] The grayscale image is subjected to full threshold segmentation using each grayscale threshold, and a corresponding segmented image under each grayscale threshold is obtained. The full threshold segmentation method belongs to the prior art and will not be described in detail here.

[0024] Step S102, for any segmented image, according to the shape characteristics of the elevator sill, each column of pixels in any segmented image is respectively organized into a longitudinal pixel sequence, and according to the gray value distribution characteristics of each longitudinal pixel sequence, the wear degree characteristic value of each longitudinal pixel sequence is correspondingly obtained, and according to the wear degree characteristic values ​​of all longitudinal pixel sequences, the local wear degree corresponding to any segmented image is obtained.

[0025] Since the segmented image under each grayscale threshold can only capture a part of the worn area in the elevator sill, the wear area in the non-light and shadow direct area can be captured at a low threshold, and the wear area in the light source direct area can be captured at a high threshold. Therefore, the degree of wear revealed in the segmented images under different grayscale thresholds is different. Therefore, in an embodiment of the present invention, each segmented image is first analyzed separately to analyze the wear condition that each segmented image can reflect.

[0026] Specifically, taking any segmented image as an example, when the grayscale threshold is higher than a certain level, the non-light source direct threshold will form a shadow background due to the grayscale intersection, which is easy to be connected with the worn area of ​​the sill surface with the same low grayscale in the light and shadow direct area, such as Figure 4Region 1 and region 2 in the image are both at the junction of the shadow background and the low grayscale region, which makes it difficult to segment the worn region. Therefore, in an embodiment of the present invention, according to the shape features of the elevator sill (vertically perpendicular to the upper and lower edge lines of the segmented image, and longitudinally parallel to the left and right edge lines of the segmented image), each column of pixels in any segmented image is respectively composed of a longitudinal pixel sequence to obtain a longitudinal pixel sequence in any segmented image, wherein the length of the longitudinal pixel sequence is consistent with the length of the segmented image (the number of pixels on the left edge line or the right edge line), and the number of the longitudinal pixel sequences is consistent with the width of the segmented image (the number of pixels on the upper edge line or the lower edge line).

[0027] After all the longitudinal pixel point sequences in any segmented image are obtained, each longitudinal pixel point sequence is analyzed separately according to the gray value distribution characteristics of each longitudinal pixel point sequence, so as to obtain the wear degree of any segmented image by judging whether the pixel point distribution on each longitudinal pixel point sequence conforms to the distribution characteristics of the surface wear of the elevator sill. The specific analysis process is as follows: The wear on the surface of the elevator sill is usually caused by the sliding of the slider or the stepping of passengers or moving objects. When a piece of wear occurs, the sliding of the slider or the friction of the passengers will produce greater friction resistance in the easily worn place, and it is easier to produce new wear around it. Therefore, this type of wear patch is usually extensible. When a piece of wear occurs, the surrounding area will be affected, thereby expanding the wear area and forming a larger patch connection domain. Mapped in the longitudinal pixel sequence, if the longitudinal pixel sequence is on the normal sill surface without wear, the low-grayscale pixels in the longitudinal pixel sequence will be evenly distributed, and there are basically no bright pixels in the longitudinal pixel sequence; if the longitudinal pixel sequence is in the shadow area, there will be more low-grayscale pixels distributed in the form of noise points, so if the longitudinal pixel sequence is in the shadow noise area, there will be more discrete low-grayscale pixels in the longitudinal pixel sequence; and if the longitudinal pixel sequence is in the worn area, the pixel distribution should be concentrated with low-grayscale pixels (the wear patch is extensible, so it will be concentrated, and the discrete distribution of low-grayscale pixels in the shadow background area can be excluded).

[0028] On the other hand, wear and tear can easily lead to vacancies inside the surface patches and jagged edges, so there are many bright pixels between the clustered low-grayscale pixels in the pixel distribution of the longitudinal pixel sequence, and bright pixels are likely to appear frequently between the clustered low-grayscale pixels. Therefore, based on the characteristics of the above two parts, each longitudinal pixel sequence is analyzed.

[0029] For any longitudinal pixel point sequence, pixels with continuous grayscale values ​​of 0 in any longitudinal pixel point sequence are grouped into pixel point sets to obtain at least two pixel point sets, and the number of interval pixel points between each two adjacent pixel point sets is obtained according to the position of the last pixel point in each of the pixel point sets in any longitudinal pixel point sequence to obtain the total number of interval pixel points and the variance of the number of interval pixel points; The number of pixels in each pixel point set is multiplied to obtain a cumulative value, the cumulative value is normalized to obtain a first normalized value, the product between the total number of interval pixels and the variance of the number of interval pixels is normalized to obtain a second normalized value, and the wear degree characteristic value of any longitudinal pixel point sequence is obtained based on the sum of the first normalized value and the second normalized value.

[0030] Among them, the calculation expression of the wear degree characteristic value of any longitudinal pixel point sequence is:

[0031] in, represents the wear degree characteristic value of the qth longitudinal pixel sequence, represents the normalization function, represents the number of pixel points in the qth vertical pixel point sequence, represents the variance of the number of pixels between each other, Indicates the total number of interval pixels.

[0032] It should be noted that It is used to characterize the accumulation of all pixel points in all pixel point sets. The more pixels there are in each pixel point set, the more concentrated the overall low-grayscale pixels are, and the more consistent with the characteristic distribution of the wear area; the more interval pixels between adjacent pixel point sets and the larger the variance, the more high-brightness pixels there are between the low-grayscale pixels in the qth longitudinal pixel point sequence, the more discrete the high-brightness pixels are, the more consistent with the distribution characteristics of the wear catfish, and the larger the wear degree characteristic value of the corresponding qth longitudinal pixel point sequence.

[0033] Similarly, the wear degree characteristic value of each longitudinal pixel point sequence is obtained. Since the wear degree characteristic value is obtained based on normalization, and after normalization, the longitudinal pixel point sequence with wear characteristic distribution, the longitudinal pixel point sequence in the normal area, and the longitudinal pixel point sequence in the shadow noise area will be better divided near the two end points of the [0, 1] interval. The wear degree characteristic value of the longitudinal pixel point sequence with wear characteristic distribution is distributed near 1, and the wear degree characteristic value of the longitudinal pixel point sequence in the normal uniform area and the shadow noise area is distributed near 0. Because the wear degree characteristic value is obtained through the normalization results of the two parts, the middle value 0.5 in the [0, 1] interval is used as the threshold of each part, and then the threshold of the wear degree characteristic value is known to be [0, 2]. The wear degree characteristic value threshold is set to 1 to divide the longitudinal pixel point sequence with wear characteristics from the longitudinal pixel point sequence without wear characteristics.

[0034] Furthermore, a wear degree characteristic value threshold is obtained, and a longitudinal pixel point sequence whose wear degree characteristic value is greater than or equal to the wear degree characteristic value threshold is used as a wear characteristic sequence, and all wear characteristic sequences in any segmented image are obtained. After all wear characteristic sequences are obtained, the local wear degree reflected in any segmented image can be analyzed by referring to the distribution similarity between the wear characteristic sequences.

[0035] Since the wear patches of the elevator sill have ductile characteristics, the wear area corresponding to the elevator sill gradually extends and expands rather than suddenly. Therefore, the distribution between adjacent longitudinal pixel point sequences should be similar. Then, a weight is set for each wear feature sequence according to the wear feature similarity between the wear feature sequence and its adjacent longitudinal pixel point sequence. At the same time, the number of pixels with a gray value of 0 in each wear feature sequence is used as the length of the wear area to obtain the local wear degree of any segmented image.

[0036] Specifically, for any wear feature sequence, the left adjacent longitudinal pixel point sequence and the right adjacent longitudinal pixel point sequence of the any wear feature sequence are obtained from all longitudinal pixel point sequences, the absolute values ​​of the differences in the wear degree characteristic values ​​between the any wear feature sequence and the left adjacent longitudinal pixel point sequence and the right adjacent longitudinal pixel point sequence are obtained, all the absolute values ​​of the differences in the wear degree characteristic values ​​are accumulated to obtain an accumulated value, and the inverse of the accumulated value is used as the wear feature similarity of the any wear feature sequence.

[0037] Among them, the calculation expression of the wear feature similarity of any wear feature sequence is:

[0038] in, represents the wear feature similarity of the i-th wear feature sequence, represents the wear degree characteristic value of the i-th wear characteristic sequence, represents the wear degree characteristic value of the longitudinal pixel point sequence adjacent to the left of the i-th wear feature sequence, represents the wear degree characteristic value of the longitudinal pixel point sequence right adjacent to the i-th wear feature sequence, | | represents the absolute value sign, and 1 represents a constant.

[0039] It should be noted that The larger the value of is, the greater the difference in wear characteristics between the i-th wear feature sequence and its left adjacent longitudinal pixel point sequence is, and the i-th wear feature sequence has less wear ductility characteristics. Similarly, The larger the value of , the greater the difference in wear features between the i-th wear feature sequence and its right adjacent longitudinal pixel point sequence, the less wear ductility characteristics the i-th wear feature sequence has, the lower the similarity between the i-th wear feature sequence and its adjacent longitudinal pixel point sequence, the lower the wear credibility of the i-th wear feature sequence, and the smaller the corresponding weight.

[0040] Similarly, the wear feature similarity of each of the wear feature sequences is obtained, and then the number of pixels with a grayscale value of 0 in each of the wear feature sequences is counted. According to the wear feature similarity of each of the wear feature sequences, a weight is set for each of the wear feature sequences. The greater the wear feature similarity, the greater the weight. Preferably, in an embodiment of the present invention, the wear feature similarity of each of the wear feature sequences is used as the weight of each of the wear feature sequences. Finally, the number of pixels of all wear feature sequences is weighted and summed to obtain the local wear degree corresponding to any segmented image.

[0041] Among them, the calculation expression of the local wear degree corresponding to any segmented image is:

[0042] in, represents the local wear degree corresponding to any segmented image, Y represents the number of wear feature sequences, represents the number of pixels with gray value 0 in the y-th wear feature sequence, Represents the weight of the yth wear feature sequence.

[0043] It should be noted that, the more pixels with gray value of 0, the higher the wear degree of the wear area composed of the wear feature sequence, and the greater the local wear degree corresponding to any segmented image.

[0044] At this point, the local wear degree corresponding to each segmented image can be obtained.

[0045] Step S103, obtaining the local wear degree corresponding to each of the segmented images, and obtaining the global wear degree of the elevator sill in the grayscale image according to the local wear degrees corresponding to all the segmented images; and determining the surface wear result of the elevator sill according to the global wear degree.

[0046] According to the distribution of the longitudinal pixel point sequences under multiple grayscale thresholds, if more grayscale thresholds have wear anomalies in the same longitudinal pixel point sequence, it means that the corresponding wear area involves a larger wear distribution range (that is, the number of segmented images involved) and a stronger wear feature. Therefore, while combining the local wear degree of all segmented images, the wear feature range is also introduced to obtain a more accurate global wear degree of the elevator sill.

[0047] Specifically, for any wear feature sequence of any segmented image, the position of any wear feature sequence in any segmented image is taken as the target position, and the number of other segmented images whose longitudinal pixel point sequence at the target position in each other segmented image is the wear feature sequence is counted and recorded as the same number; Obtain the same number corresponding to each wear feature sequence in any of the segmented images, obtain the total number of the same number of any of the segmented images, obtain the total number of the same number of each of the segmented images, accumulate the product of the local wear degree of all segmented images and the total number of the same number, and obtain the global wear degree of the elevator sill in the grayscale image.

[0048] Among them, the calculation expression of the global wear degree of the elevator sill is:

[0049] in, represents the global wear degree of the elevator sill, L represents the number of segmented images, represents the local wear degree corresponding to the i-th segmented image, Represents the total amount of the same number corresponding to the i-th segmented image.

[0050] It should be noted that The larger the value is, the larger the wear distribution is in the same wear area, and the stronger the global wear of the entire elevator sill is.

[0051] After obtaining the global wear degree of the elevator sill in the grayscale image, the surface wear result of the elevator sill can be determined according to the global wear degree. Specifically, the global wear degree of the elevator sill is equally divided into evaluation levels using a linear normalization function norm(). First, the global wear degree is normalized to obtain a normalized wear value. ,when When the elevator sill surface is considered to have no wear; when When the elevator sill surface is slightly worn, it is considered that there is no impact on the normal use of the elevator; when When the elevator sill surface is considered to be worn and will affect the normal use of the elevator, the elevator maintenance personnel will be notified remotely to perform maintenance and repairs on the elevator within one working day; At this time, it is considered that there is strong and large-scale wear on the surface of the elevator sill, which has had a serious impact on the daily use of the elevator and poses a major safety hazard. It is necessary to immediately notify the elevator maintenance personnel remotely, shut down the elevator and conduct maintenance and repairs on the elevator.

[0052] It should be noted that 0.75 is a severe wear value threshold, that is, when the normalized wear value is greater than or equal to the preset severe wear value threshold, it is determined that the surface of the elevator sill is severely worn.

[0053] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A method for identifying the degree of wear on the surface of an elevator sill based on computer vision, characterized in that: The method comprises: Acquire a surface image of an elevator sill, wherein the direction of the elevator sill in the surface image is parallel to the longitudinal edge of the image, grayscale the surface image to obtain a corresponding grayscale image; perform full threshold segmentation on the grayscale image using at least two preset grayscale thresholds, and obtain at least two segmented images accordingly; For any segmented image, according to the shape characteristics of the elevator sill, each column of pixels in the segmented image is respectively formed into a longitudinal pixel sequence, and according to the gray value distribution characteristics of each longitudinal pixel sequence, the wear degree characteristic value of each longitudinal pixel sequence is correspondingly obtained, and according to the wear degree characteristic values ​​of all longitudinal pixel sequences, the local wear degree corresponding to any segmented image is obtained; The local wear degree corresponding to each of the segmented images is obtained, and the global wear degree of the elevator sill in the grayscale image is obtained according to the local wear degrees corresponding to all the segmented images; and the surface wear result of the elevator sill is determined according to the global wear degree.

2. The method for identifying the degree of wear of elevator sill surface based on computer vision according to claim 1, characterized in that: The full threshold segmentation means that the grayscale value greater than the grayscale threshold is set to 255, and the grayscale value less than or equal to the grayscale threshold is set to 0. Then, according to the grayscale value distribution characteristics of each longitudinal pixel point sequence, the wear degree characteristic value of each longitudinal pixel point sequence is correspondingly obtained, including: For any longitudinal pixel point sequence, pixels with continuous grayscale values ​​of 0 in any longitudinal pixel point sequence are grouped into pixel point sets to obtain at least two pixel point sets, and the number of interval pixel points between each two adjacent pixel point sets is obtained according to the position of the last pixel point in each of the pixel point sets in any longitudinal pixel point sequence to obtain the total number of interval pixel points and the variance of the number of interval pixel points; The number of pixels in each pixel point set is multiplied to obtain a cumulative value, the cumulative value is normalized to obtain a first normalized value, the product between the total number of interval pixels and the variance of the number of interval pixels is normalized to obtain a second normalized value, and the wear degree characteristic value of any longitudinal pixel point sequence is obtained based on the sum of the first normalized value and the second normalized value.

3. The method for identifying the degree of wear of elevator sill surface based on computer vision according to claim 2, characterized in that: The obtaining of the local wear degree corresponding to any segmented image according to the wear degree characteristic values ​​of all longitudinal pixel point sequences includes: Obtaining a preset wear degree characteristic value threshold, and taking a longitudinal pixel point sequence whose wear degree characteristic value is greater than or equal to the wear degree characteristic value threshold as a wear characteristic sequence; The wear feature similarity of each wear feature sequence is obtained, the number of pixels with a gray value of 0 in each wear feature sequence is counted, and the local wear degree corresponding to any segmented image is obtained according to the wear feature similarity and the number of pixels of each wear feature sequence.

4. The method for identifying the degree of wear of elevator sill surface based on computer vision according to claim 3, characterized in that: The obtaining of the wear feature similarity of each wear feature sequence comprises: For any wear feature sequence, obtain the left adjacent longitudinal pixel point sequence and the right adjacent longitudinal pixel point sequence of the any wear feature sequence from all longitudinal pixel point sequences, obtain the absolute values ​​of the wear degree characteristic value differences between the any wear feature sequence and the left adjacent longitudinal pixel point sequence and the right adjacent longitudinal pixel point sequence, respectively, accumulate all the absolute values ​​of the wear degree characteristic value differences to obtain the accumulated value, and use the inverse of the accumulated value as the wear feature similarity of the any wear feature sequence.

5. The method for identifying the degree of wear of elevator sill surface based on computer vision according to claim 3, characterized in that: The obtaining of the local wear degree corresponding to any segmented image according to the wear feature similarity and the number of pixels of each wear feature sequence includes: According to the wear feature similarity of each wear feature sequence, a weight is set for each wear feature sequence. The greater the wear feature similarity, the greater the weight. The number of pixel points of all wear feature sequences is weighted and summed to obtain the local wear degree corresponding to any segmented image.

6. The method for identifying the degree of wear of elevator sill surface based on computer vision according to claim 3, characterized in that: The step of obtaining the global wear degree of the elevator sill in the grayscale image according to the local wear degrees corresponding to all the segmented images includes: For any wear feature sequence of any segmented image, the position of any wear feature sequence in any segmented image is taken as the target position, and the number of other segmented images whose longitudinal pixel point sequence at the target position in each other segmented image is the wear feature sequence is counted and recorded as the same number; Obtain the same number corresponding to each wear feature sequence in any of the segmented images, obtain the total number of the same number of any of the segmented images, obtain the total number of the same number of each of the segmented images, accumulate the product of the local wear degree of all segmented images and the total number of the same number, and obtain the global wear degree of the elevator sill in the grayscale image.

7. The method for identifying the degree of wear of elevator sill surface based on computer vision according to claim 1, characterized in that: The determining the surface wear result of the elevator sill according to the global wear degree includes: The global wear degree is normalized to obtain a normalized wear value, and when the normalized wear value is greater than or equal to a preset severe wear value threshold, it is determined that the surface of the elevator sill is severely worn.

Images