CN115205297B - Abnormal state detection method for pneumatic winch - Google Patents
Abnormal state detection method for pneumatic winch Download PDFInfo
- Publication number
- CN115205297B CN115205297B CN202211133868.XA CN202211133868A CN115205297B CN 115205297 B CN115205297 B CN 115205297B CN 202211133868 A CN202211133868 A CN 202211133868A CN 115205297 B CN115205297 B CN 115205297B
- Authority
- CN
- China
- Prior art keywords
- gradient
- steel wire
- wire rope
- frequency
- amplitude
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 27
- 238000001514 detection method Methods 0.000 title claims abstract description 14
- 229910000831 Steel Inorganic materials 0.000 claims abstract description 142
- 239000010959 steel Substances 0.000 claims abstract description 142
- 238000005315 distribution function Methods 0.000 claims abstract description 42
- 238000000034 method Methods 0.000 claims abstract description 35
- 230000005856 abnormality Effects 0.000 claims abstract description 6
- 238000012937 correction Methods 0.000 claims description 16
- 230000001186 cumulative effect Effects 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 14
- 238000013528 artificial neural network Methods 0.000 claims description 10
- 150000001875 compounds Chemical class 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 230000009467 reduction Effects 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003638 chemical reducing agent Substances 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000004804 winding Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- G06T5/90—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an abnormal state detection method of a pneumatic winch, belonging to the technical field of intelligent winch detection; the method comprises the following steps: acquiring a steel wire rope gray scale image; obtaining a plurality of different frequency threshold distribution functions; respectively correcting the gradient amplitude frequency according to the frequency threshold value calculated by each frequency threshold value distribution function, and equalizing the gradient histogram by using the corrected gradient amplitude frequency to obtain a plurality of equalized gradient histograms; obtaining a clear steel wire rope image by calculating the gradient amplitude entropy of each equalized gradient histogram; and detecting the abnormality of the steel wire rope according to the obtained clear steel wire rope image. According to the invention, the acquired steel wire rope image is processed by using computer vision, the steel wire rope image is subjected to image enhancement by combining the steel wire rope gradient histogram, and the steel wire rope abnormity is acquired according to the enhanced image.
Description
Technical Field
The invention relates to the technical field of winch intelligent detection, in particular to an abnormal state detection method of a pneumatic winch.
Background
The pneumatic winch takes the blade type pneumatic motor as power, transmits the power to the roller through the planetary gear reducer to lift heavy objects, and has the characteristics of small volume, light weight, compact structure, simple operation, convenient maintenance and the like. Before the pneumatic winch is used, the state of the pneumatic winch needs to be checked so as to ensure the safety in the using process.
The steel wire rope of the pneumatic winch is an important part for lifting heavy objects, so that the steel wire rope is of great importance for the abnormal inspection of the steel wire rope. The steel wire rope in the gear box can be inspected for abnormity manually, but the steel wire rope above the pneumatic winch is easy to be inspected by mistake or missed because the steel wire rope is far away from an operator.
At present, in the process of detecting the steel wire rope through image processing, if abnormity detection is directly carried out according to the steel wire rope image with poor definition, the accuracy rate of the obtained result is low. Therefore, the gray scale image of the steel wire rope needs to be enhanced. The existing image enhancement method such as histogram equalization has the problems of image entropy reduction, detail loss and the like, and the gradient histogram equalization has the problem of excessive enhancement.
Disclosure of Invention
In order to solve the defects in the prior art, the invention provides a method for detecting the abnormal state of a pneumatic winch, which utilizes computer vision to process the acquired steel wire rope image, and firstly, a frequency threshold value distribution function is obtained according to the distribution of gradient amplitudes in a steel wire rope gradient histogram; secondly, respectively correcting gradient amplitude frequencies according to frequency thresholds calculated by a frequency threshold distribution function to obtain an equalized gradient histogram with the maximum gradient amplitude entropy; then, enhancing the steel wire rope gray level image by using the obtained equalized gradient histogram to obtain a clear steel wire rope image; finally, carrying out abnormity detection on the steel wire rope according to the enhanced clear steel wire rope image to obtain the abnormity of the steel wire rope; therefore, the abnormity of the steel wire rope above the pneumatic winch is checked, and meanwhile, the abnormity of the steel wire rope in the gear box can be checked, so that the checking accuracy and the checking efficiency are improved.
The invention aims to provide an abnormal state detection method of a pneumatic winch, which comprises the following steps:
acquiring a steel wire rope gray scale image; acquiring a steel wire rope gradient histogram according to the gradient amplitude of each pixel point in the steel wire rope gray scale map; acquiring a boundary gradient amplitude according to the accumulated frequency of each gradient amplitude in the steel wire rope gradient histogram;
presetting a frequency threshold distribution function according to the distribution of gradient amplitudes in the gradient histogram, and acquiring a plurality of different frequency threshold distribution functions according to the gradient amplitude frequencies in the gradient histogram and the boundary gradient amplitudes;
respectively correcting the gradient amplitude frequency according to the frequency threshold value calculated by each frequency threshold value distribution function, and equalizing the gradient histogram by using the corrected gradient amplitude frequency to obtain a plurality of equalized gradient histograms; acquiring an equalized gradient histogram with the maximum gradient amplitude entropy by calculating the gradient amplitude entropy of each equalized gradient histogram; enhancing the steel wire rope gray level image by using the obtained equalized gradient histogram to obtain a clear steel wire rope image;
and detecting the abnormality of the steel wire rope according to the obtained clear steel wire rope image.
Preferably, the frequency threshold distribution function is as follows:
in the formula (I), the compound is shown in the specification,、、the amplitude, standard deviation and mean parameter of the frequency threshold distribution function;
More preferably, the gradient amplitude frequency and the boundary gradient amplitude in the gradient histogram are respectively obtainedThe value range of (a).
Wherein, theIs the maximum value of gradient amplitude;is the demarcation gradient magnitude;for a gradient histogram corresponding to 0Gradient amplitude frequency of gradient amplitude.
More preferably, a plurality of different said frequency threshold distribution functions are parametric、With 0.1 as step length in the value range of (A), traversing to obtain parameters、At a parameterIn the value range of (1) and using 1 as step length to traverse and obtain parametersAnd then acquired.
More preferably, the constraints of the frequency threshold distribution function are obtained by using the equalized features:
wherein the content of the first and second substances,to demarcate gradient amplitudesThe correction frequency of (1);the corrected frequency mean value is obtained;is the gradient amplitudeThe correction frequency of (1);is the smallest gradient amplitudeA value;the maximum gradient magnitude;is the demarcation gradient magnitude.
wherein the content of the first and second substances,is the gradient amplitudeA frequency threshold of (a);to the corrected frequency mean:is the gradient amplitudeOf (c) is detected.
in the formula (I), the compound is shown in the specification,is the gradient amplitude;the maximum gradient magnitude;the number of gradient magnitudes.
Preferably, the step of obtaining the boundary gradient magnitude is as follows:
obtaining the ratio of the internal pixel points of the steel wire to all the pixel points of the steel wire in the steel wire rope;
Acquiring the cumulative frequency of each gradient amplitude of the gradient histogram, wherein when the cumulative frequency of one gradient amplitude is not less thanAnd the cumulative frequency of the gradient amplitude preceding the gradient amplitude <Then, the gradient amplitude is used as the boundary gradient amplitude.
Preferably, the process of detecting the abnormality of the wire rope includes the steps of:
and establishing a neural network, inputting a clear steel wire rope image into the neural network, and outputting the abnormal type, the abnormal position and the abnormal size of the steel wire rope.
The invention has the beneficial effects that:
the invention provides a method for detecting an abnormal state of a pneumatic winch. And secondly, combining the gradient histogram of the steel wire rope to perform image enhancement on the steel wire rope image. And finally, acquiring the abnormity of the steel wire rope according to the enhanced steel wire rope image.
The method mainly utilizes computer vision to process the acquired steel wire rope image, and combines the steel wire rope gradient histogram to perform image enhancement on the steel wire rope image. And acquiring the abnormity of the steel wire rope according to the enhanced image.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart illustrating the general steps of an embodiment of a method for detecting an abnormal condition of a pneumatic winch according to the present invention;
fig. 2 is a gradient histogram of a steel cord.
Fig. 3 is a schematic diagram of a frequency threshold distribution function.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The specific scenes aimed by the invention are as follows: pneumatic winches need to be inspected before they are used. According to the invention, the steel wire rope is enhanced by shooting the steel wire rope image, and the steel wire rope abnormity is obtained according to the enhanced image. The pneumatic winch is safe to use. Meanwhile, the invention can also be applied to the pneumatic winch for checking the abnormity of the steel wire rope in the using process, and the pneumatic winch can be immediately stopped if the abnormity occurs in the using process.
The invention provides a method for detecting an abnormal state of a pneumatic winch, which is shown in figure 1 and comprises the following steps:
s1, acquiring a steel wire rope image, and performing gray level processing to obtain a steel wire rope gray level image; acquiring a gradient histogram of the steel wire rope according to the gradient amplitude of each pixel point in the gray scale map of the steel wire rope; acquiring a boundary gradient amplitude according to the accumulated frequency of each gradient amplitude in the steel wire rope gradient histogram;
in the process of acquiring the steel wire rope image, a camera is erected on the side edge of the steel wire rope above the pneumatic winch, and the initial image of the steel wire rope above the pneumatic winch is shot. Or a camera is erected above the pneumatic winch gear box to shoot an initial image of the steel wire rope in the pneumatic winch gear box. The image includes a background and a wire rope.
Identifying the target in the segmented image by adopting a DNN semantic segmentation mode for the acquired initial image; the relevant content of the DNN network is as follows: the dataset used was the wire rope initial image dataset. The pixels to be segmented are divided into 2 types, namely the labeling process of the training set corresponding to the labels is as follows: and in the semantic label of the single channel, the label of the pixel at the corresponding position belonging to the background class is 0, and the label of the pixel belonging to the steel wire rope is 1. The task of the network is classification, so the loss function used is a cross entropy loss function.
The method comprises the steps of processing a steel wire rope initial image through DNN, obtaining a steel wire rope shade in the image, multiplying the steel wire rope shade with the steel wire rope initial image, removing a background, and finally obtaining a steel wire rope image, wherein the steel wire rope image only contains steel wire ropes. For convenience of analysis, the steel wire rope image is subjected to gray processing to obtain a steel wire rope gray image.
In order to obtain a larger visual field as much as possible, the camera positioned on the side edge of the steel wire rope above the pneumatic winch is far away from the steel wire rope and is influenced by the environment and weather, and the definition of a shot steel wire rope image is poor. If the abnormity detection is directly carried out according to the steel wire rope image with poor definition, the accuracy rate of the obtained result is low. Therefore, the gray scale image of the steel wire rope needs to be enhanced.
The existing image enhancement method such as histogram equalization has the problems of image entropy reduction, detail loss and the like, and the gradient histogram equalization has the problem of over-enhancement.
According to the invention, the frequency threshold distribution function is obtained, the threshold segmentation is carried out on the gradient histogram, and the gradient histogram equalization is carried out according to the segmentation result to realize the image enhancement. The method can ensure that the entropy of the image is not reduced, prevent the excessive enhancement and enhance the details of the image.
And (3) an image enhancement process:
s2, acquiring a gradient histogram of the steel wire rope according to the gradient amplitude of each pixel point in the gray-scale map of the steel wire rope; the gradient amplitude of each pixel point in the steel wire rope gray scale image is obtained mainly by utilizing a Sobel operator. Drawing a steel wire rope gradient histogram by taking the gradient amplitude as a horizontal axis and the frequency of the gradient amplitude corresponding to the steel wire rope gray scale map as a vertical axis; as shown in fig. 2.
Acquiring a boundary gradient amplitude according to the accumulated frequency of each gradient amplitude in the steel wire rope gradient histogram; the steel wire rope is formed by twisting steel wire bundles, and the steel wire bundles are formed by twisting steel wires. Therefore, the inner part of the steel wire is flat in the steel wire rope image, and the gradient amplitude is smaller. The boundary gradient magnitude is thus obtained such that the gradient magnitude distribution on the left side of the boundary gradient magnitude is unchanged, such that the gradient magnitude distribution on the right side of the boundary gradient magnitude is changed. Obtaining the ratio of the pixel points in the steel wire to all the pixel points of the steel wire according to the historical experience value(in this case, the). Obtaining the cumulative frequency of each gradient amplitude of the gradient histogram, when the cumulative frequency of one gradient amplitude is larger than or equal toAnd the cumulative frequency of the previous gradient amplitude is less thanThen, the gradient amplitude is taken as the boundary gradient amplitude and recorded as. Focusing off in image enhancementNote the distribution of the boundary gradient amplitude and the gradient amplitude on the right side of it with the greater frequency.
S3, presetting a frequency threshold distribution function according to the distribution of gradient amplitudes in the gradient histogram, and acquiring a plurality of different frequency threshold distribution functions according to gradient amplitude frequencies and boundary gradient amplitudes in the gradient histogram;
the existing gradient histogram equalization method realizes image enhancement by changing the distribution of gradient amplitudes in a gradient histogram by using an equalization means. The frequency of the non-edge part of the image is large, the gradient amplitude of the pixel points of the non-edge part is small, the frequency of the edge part of the image is small, and the gradient of the pixel points of the edge part is large. Therefore, in the gradient histogram, the non-edge portion is always large in frequency and small in gradient, and the edge portion is small in frequency and large in gradient. After equalization, the non-edge part with high frequency may press the edge part with low frequency during equalization, so that the distribution of the enhanced non-edge part occupies an excessively large range of gradient amplitude, and further the target gradient histogram may have a degradation effect, such as excessive gradient enhancement.
For the steel wire rope image, the pixel point of the gradient amplitude on the left side of the gradient histogram (namely the pixel point with the smaller gradient amplitude) is a flat area on the image and may be the pixel point inside the steel wire forming the steel wire rope, so that the distribution of the gradient amplitude on the left side of the gradient histogram does not need to be changed too much, and the over-enhancement is avoided. The pixel point of the gradient threshold value on the right side of the gradient histogram (namely the pixel point with larger gradient amplitude) has larger difference with the surrounding pixel points on the image and is clearer, so that the gradient amplitude distribution on the right side of the gradient histogram does not need to be changed too much. The pixel point with the gradient amplitude in the middle of the gradient histogram may be the characteristics of edge, defect and the like on the image, but the difference between the pixel point and the surrounding pixel points is small and unclear, so the distribution of the gradient amplitude in the middle of the gradient histogram needs to be focused in the image enhancement.
The equalization operation actually widens the gradient amplitude range with a larger frequency, and the widening degree is larger for the gradient amplitude range with a larger frequency. In order to implement the above-mentioned "not excessively changing the gradient amplitude distribution on the left side of the gradient histogram, not excessively changing the gradient amplitude distribution on the right side of the gradient histogram, focusing on the gradient amplitude distribution in the middle of the gradient histogram" by using the gradient histogram equalization operation, a frequency threshold may be set for the gradient histogram, a smaller frequency threshold may be set for the gradient amplitudes on the left and right sides of the gradient histogram, and a larger frequency threshold may be set for the gradient amplitude in the middle of the gradient histogram. And for the gradient amplitude with the frequency larger than the frequency threshold, carrying out equalization according to the frequency threshold, and for the gradient amplitude with the frequency smaller than the frequency threshold, carrying out equalization according to the frequency. The gaussian distribution is bell-shaped, low at both ends and high in the middle, and meets the setting requirement of the above frequency threshold, so, referring to fig. 3, the gaussian distribution probability density function can be preset as the frequency threshold distribution function:
in the formula (I), the compound is shown in the specification,、the amplitude, standard deviation and mean value parameters of the frequency threshold distribution function are obtained;is the gradient amplitudeIs detected.
For a frequency threshold distribution function (Gaussian distribution), the random variable falls onThe probability is very small, the corresponding event is not considered to happen in practical problem, and the interval can be basically setThe frequency threshold distribution function is regarded as a practically possible value interval of the random variable, and all gradient amplitudes need to be covered, so that all gradient amplitudes are required to be located in the interval, the minimum value of the gradient amplitudes is 0, and the maximum value is 0The length of the gradient amplitude range is. Parameter at mean valueOn the left, the frequency threshold distribution function increases monotonically, at the mean parameterOn the right, the frequency threshold distribution function decreases monotonically. In order to focus on the distribution of the boundary gradient amplitude and the gradient amplitude with a higher frequency on the right side thereof in image enhancement, the boundary gradient amplitude is requiredIs greater than the threshold of the gradient amplitude to its left, thenShould be in a range where the frequency threshold distribution function monotonically increases, then. In order for the frequency threshold distribution function to be meaningful, its maximum value (i.e., its maximum value)Time value) needs to be greater than 0 and less than the maximum gradient amplitude frequency. Gradient amplitude frequency in the gradient histogram is recorded as. According to the above parameters、、The constraint conditions of (1) are:
Wherein the content of the first and second substances,is the maximum value of gradient amplitude;dividing gradient amplitude;for a gradient histogram corresponding to 0Gradient amplitude frequency of gradient amplitude.
According to gradient amplitude frequency in gradient histogram and boundary gradient amplitude to frequency threshold distribution functionThe value of,Is taken from andto obtain a plurality of different frequency threshold distribution functions; in particular, parameters are determined、With 0.1 as step length in the value range of (A), traversing to obtain parameters、At a parameterThe parameter is obtained by traversing with 1 as step length in the value range of (A) to obtain different frequency threshold distribution functions.
S4, respectively correcting the gradient amplitude frequency according to the frequency threshold value calculated by each frequency threshold value distribution function, and equalizing the gradient histogram by using the corrected gradient amplitude frequency to obtain a plurality of equalized gradient histograms;
modifying the gradient amplitude frequency according to a frequency threshold distribution function, the gradient amplitudeIs the corrected frequencyComprises the following steps:
wherein the content of the first and second substances,is the gradient amplitudeA frequency threshold of (a);to the corrected frequency mean:is the gradient amplitudeOf (c) is detected.
in the formula (I), the compound is shown in the specification,is the gradient amplitude;the maximum gradient magnitude;the number of gradient magnitudes.
The equalization operation maps the gradient amplitude to the cumulative correction frequency of the gradient amplitudeWithin the interval, when the difference between the cumulative correction frequency of the gradient amplitude and the cumulative correction frequency of the previous gradient amplitude (i.e. the correction frequency of the gradient amplitude) is smaller than the corrected frequency average valueThen, the gradient amplitude and the previous gradient amplitude are mapped to the same position, so that the gradient amplitude and the previous gradient amplitude are combined; when the difference between the cumulative correction frequency of the gradient amplitude and the cumulative correction frequency of the previous gradient amplitude (i.e. the correction frequency of the gradient amplitude) is greater than or equal to the corrected frequency mean valueAnd is less thanWhen the gradient amplitude is mapped to the position behind the previous gradient amplitude mapping position, the relative position relationship between the gradient amplitude and the previous gradient amplitude after mapping is unchanged; when the difference between the cumulative correction frequency of the gradient amplitude and the cumulative correction frequency of the previous gradient amplitude (i.e. the correction frequency of the gradient amplitude) is greater than or equal to the corrected frequency mean valueWhen the gradient amplitude and the previous gradient amplitude are mapped to different positions, and a gap exists between the gradient amplitude and the previous gradient amplitude, the relative position of the gradient amplitude and the previous gradient amplitude after mapping is changed.
Acquiring the constraint conditions of the frequency threshold distribution function by using the equalized characteristics:
wherein the content of the first and second substances,to demarcate gradient amplitudesThe correction frequency of (2);the corrected frequency mean value is obtained;is the gradient amplitudeThe correction frequency of (2);the minimum gradient magnitude;the maximum gradient magnitude;is the demarcation gradient magnitude.
To make itThe gradient amplitude distribution within the range is unchanged after equalization, then the requirement is thatGradient amplitude frequency in the range of greater than or equal toAnd is less than(ii) a To change the boundary gradient amplitude and the gradient amplitude distribution on the right side thereof, the frequency of the boundary gradient amplitude is required to be equal to or higher than。
It should be noted that, the gradient amplitude frequency and the boundary gradient amplitude in the gradient histogram are used as the threshold distribution function of the frequencyThe value of,The values and the values of (a) are limited to obtain a plurality of different frequency threshold distribution functions; the method comprises the steps of obtaining a plurality of frequency threshold distribution functions, wherein the frequency threshold distribution functions meeting the constraint conditions of the frequency threshold distribution functions are multiple, the gradient amplitude frequency is corrected according to each frequency threshold distribution function, the gradient histogram is equalized by the corrected gradient amplitude frequency, and a plurality of equalized gradient histograms are obtained.
S5, obtaining the equalized gradient histogram with the maximum gradient amplitude entropy by calculating the gradient amplitude entropy of each equalized gradient histogram; enhancing the steel wire rope gray level image by using the obtained equalized gradient histogram to obtain a clear steel wire rope image;
calculating gradient magnitude entropy of each equalized gradient histogram from the obtained equalized gradient histograms, such asEntropy of gradient magnitude of equalized gradient histogramComprises the following steps:
wherein the content of the first and second substances,is the gradient amplitudeThe frequencies in the first histogram of equalized gradients,the maximum gradient magnitude;
in the equalization operation, different gradient amplitudes may be combined into one gradient amplitude, and the number of equalized gradient amplitudes is reduced, so that the gradient information amount of the image is reduced. For the equalized image, the smaller the magnitude of the reduction in the gradient information amount, the better. Therefore when the gradient magnitude entropyThe larger the equalization result, the larger the number of gradient amplitudes existing after equalization, the smaller the amplitude of gradient information amount reduction, and the better the equalization result.
And acquiring an equalization gradient histogram with the maximum gradient amplitude entropy, wherein the gradient histogram is the optimal equalization gradient histogram. And reconstructing an enhanced image according to the optimal equalized gradient histogram.
Therefore, the steel wire rope gray level image is enhanced, and a clear steel wire rope image is obtained.
S6, detecting the abnormity of the steel wire rope according to the obtained clear steel wire rope image; the method mainly detects the abnormity of the steel wire rope by an Encoder-Decoder structure through a DNN neural network. And the steel wire rope abnormity detection is carried out according to the clear steel wire rope image, so that the identification result is more accurate.
The method uses the DNN neural network to detect the abnormity of the steel wire rope by using the Encoder-Decoder structure. The method mainly refers to a training neural network to identify an abnormal process, specifically, by establishing a DNN neural network, clear steel wire rope images are input into the neural network, and abnormal types, abnormal positions and sizes of the steel wire ropes are output;
it should be noted that:
1) The DNN neural network adopts an Encoder-Decoder form, firstly encodes a clear steel wire rope image and then decodes the steel wire rope image; the input of the network is a clear steel wire rope image, and the output is a center point of a surrounding frame, the length, width, height and size of the regressed surrounding frame, an abnormal category and confidence coefficient;
2) The method comprises the steps that the input of a network is a clear steel wire rope image, the clear steel wire rope image is coded firstly, namely, the features of the clear steel wire rope image are extracted in the process of downsampling the clear steel wire rope image by using convolution and pooling operations, and the output of a coder is an extracted feature vector;
3) The input of the decoder is the output characteristic vector of the encoder, and the decoder returns the central point and the length, width and height dimensions of the target corresponding to the surrounding frame in the clear steel wire rope image through up-sampling; the output of the decoder is the output of the network;
4) The data set used for training the network comprises clear steel wire rope images;
5) The clear label of the steel wire rope image is a steel wire rope abnormal category (bending, knotting, rusting, wire breaking and winding) and a surrounding frame corresponding to the abnormal category, and comprises a surrounding frame central point coordinate and the length, width and height dimensions of the surrounding frame;
6) The loss function is a mean square error loss function;
thus, the abnormity identification of the steel wire rope is completed; the abnormal state of the steel wire rope influences the use of the pneumatic winch, and the steel wire rope is adjusted or replaced in time by combining the recognition result, so that the safety of the pneumatic winch in the use process is ensured.
In summary, according to the abnormal state detection method for the pneumatic winch provided by the invention, firstly, an image of the steel wire rope of the pneumatic winch is shot, and semantic segmentation is performed on the image to obtain the connected domain of the steel wire rope; secondly, combining the gradient histogram of the steel wire rope to perform image enhancement on the steel wire rope image; and finally, acquiring the abnormity of the steel wire rope according to the enhanced steel wire rope image.
The method mainly comprises the steps of processing an acquired steel wire rope image by using computer vision, and performing image enhancement on the steel wire rope image by combining a steel wire rope gradient histogram; and acquiring the abnormality of the steel wire rope according to the enhanced image.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent substitutions, improvements, etc. within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (5)
1. The abnormal state detection method of the pneumatic winch is characterized by comprising the following steps of:
acquiring a steel wire rope gray scale image; acquiring a gradient histogram of the steel wire rope according to the gradient amplitude of each pixel point in the gray scale map of the steel wire rope; acquiring a boundary gradient amplitude according to the accumulated frequency of each gradient amplitude in the steel wire rope gradient histogram; the method comprises the following steps of obtaining a boundary gradient amplitude value:
obtaining the ratio of the internal pixel points of the steel wire to all the pixel points of the steel wire in the steel wire rope;
Acquiring the cumulative frequency of each gradient amplitude of the gradient histogram, wherein when the cumulative frequency of one gradient amplitude is not less thanAnd the cumulative frequency of the gradient amplitude preceding the gradient amplitude <Taking the gradient amplitude as a boundary gradient amplitude;
presetting a frequency threshold distribution function according to the distribution of gradient amplitudes in the gradient histogram, and acquiring a plurality of different frequency threshold distribution functions according to the gradient amplitude frequencies in the gradient histogram and the boundary gradient amplitudes;
the frequency threshold distribution function is as follows:
in the formula (I), the compound is shown in the specification,、、the amplitude, standard deviation and mean parameter of the frequency threshold distribution function;
respectively correcting the gradient amplitude frequency according to the frequency threshold value calculated by each frequency threshold value distribution function, and equalizing the gradient histogram by using the corrected gradient amplitude frequency to obtain a plurality of equalized gradient histograms; acquiring an equalized gradient histogram with the maximum gradient amplitude entropy by calculating the gradient amplitude entropy of each equalized gradient histogram; enhancing the steel wire rope gray level image by using the obtained equalized gradient histogram to obtain a clear steel wire rope image;
wherein, the first and the second end of the pipe are connected with each other,is the gradient amplitudeA frequency threshold of (a);to the corrected frequency mean:is the gradient amplitudeThe frequency of (d);
in the formula (I), the compound is shown in the specification,is the gradient amplitude;the maximum gradient magnitude;the number of gradient amplitudes;
acquiring the constraint conditions of the frequency threshold distribution function by using the equalized characteristics:
wherein, the first and the second end of the pipe are connected with each other,to demarcate gradient amplitudesThe correction frequency of (2);is the corrected frequency mean value;is the gradient amplitudeThe correction frequency of (1);the minimum gradient magnitude;the maximum gradient magnitude;is the demarcation gradient magnitude;
and detecting the abnormality of the steel wire rope according to the obtained clear steel wire rope image.
3. The method of detecting an abnormal state of a pneumatic winch according to claim 2,
4. The pneumatic air winch of claim 2The constant state detection method is characterized in that a plurality of different frequency threshold distribution functions are in parameters、With 0.1 as step length in the value range of (A), traversing to obtain parameters、At a parameterIn the value range of (1) and using 1 as step length to traverse and obtain parametersAnd then acquired.
5. The method for detecting the abnormal state of the pneumatic winch according to claim 1, wherein the process of detecting the abnormality of the wire rope comprises the steps of:
and establishing a neural network, inputting a clear steel wire rope image into the neural network, and outputting the abnormal type, the abnormal position and the abnormal size of the steel wire rope.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211133868.XA CN115205297B (en) | 2022-09-19 | 2022-09-19 | Abnormal state detection method for pneumatic winch |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211133868.XA CN115205297B (en) | 2022-09-19 | 2022-09-19 | Abnormal state detection method for pneumatic winch |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115205297A CN115205297A (en) | 2022-10-18 |
CN115205297B true CN115205297B (en) | 2022-12-02 |
Family
ID=83572839
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211133868.XA Active CN115205297B (en) | 2022-09-19 | 2022-09-19 | Abnormal state detection method for pneumatic winch |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115205297B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115409745B (en) * | 2022-10-31 | 2023-01-31 | 深圳市亿康医疗技术有限公司 | CT image enhancement method applied to radiotherapy preparation |
CN117237357B (en) * | 2023-11-15 | 2024-01-30 | 上海杰臻电气技术有限公司 | Machine vision-based steel wire rope online monitoring system and method |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114821314A (en) * | 2022-04-19 | 2022-07-29 | 中铁建设集团有限公司 | Airport pavement abnormity detection method based on machine vision |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100278423A1 (en) * | 2009-04-30 | 2010-11-04 | Yuji Itoh | Methods and systems for contrast enhancement |
CN102222323A (en) * | 2011-06-13 | 2011-10-19 | 北京理工大学 | Histogram statistic extension and gradient filtering-based method for enhancing infrared image details |
CN112053309A (en) * | 2020-10-24 | 2020-12-08 | 北方民族大学 | Image enhancement method and image enhancement device |
CN114331925B (en) * | 2022-03-15 | 2022-07-19 | 北京锐影医疗技术有限公司 | Multi-scale relative gradient histogram equalization method and device |
CN114705686A (en) * | 2022-03-21 | 2022-07-05 | 佛山科学技术学院 | Method and system for detecting surface defects of cylindrical pot, controller and industrial personal computer |
CN114757854B (en) * | 2022-06-15 | 2022-09-02 | 深圳市安星数字系统有限公司 | Night vision image quality improving method, device and equipment based on multispectral analysis |
-
2022
- 2022-09-19 CN CN202211133868.XA patent/CN115205297B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114821314A (en) * | 2022-04-19 | 2022-07-29 | 中铁建设集团有限公司 | Airport pavement abnormity detection method based on machine vision |
Also Published As
Publication number | Publication date |
---|---|
CN115205297A (en) | 2022-10-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115205297B (en) | Abnormal state detection method for pneumatic winch | |
CN114842017B (en) | HDMI cable surface quality detection method and system | |
CN115170576B (en) | Aluminum pipe surface defect detection method based on machine vision | |
CN106683075B (en) | Method for detecting bolt defects at cross arm of power transmission line tower | |
CN111402248A (en) | Transmission line lead defect detection method based on machine vision | |
CN115294140B (en) | Hardware part defect detection method and system | |
CN111814686A (en) | Vision-based power transmission line identification and foreign matter invasion online detection method | |
CN115082481A (en) | Image enhancement method for gear | |
CN115063430B (en) | Electric pipeline crack detection method based on image processing | |
CN107742307A (en) | Based on the transmission line galloping feature extraction and parameters analysis method for improving frame difference method | |
CN110414308B (en) | Target identification method for dynamic foreign matters on power transmission line | |
CN115631116B (en) | Aircraft power inspection system based on binocular vision | |
CN115330790B (en) | Image-based cable strand quality detection method and system | |
CN116665080B (en) | Unmanned aerial vehicle deteriorated insulator detection method and system based on target recognition | |
CN116630813B (en) | Highway road surface construction quality intelligent detection system | |
CN114926407A (en) | Steel surface defect detection system based on deep learning | |
CN113313107A (en) | Intelligent detection and identification method for multiple types of diseases on cable surface of cable-stayed bridge | |
CN107273884A (en) | A kind of License Plate Identification method based on mobile terminal camera | |
CN115330796A (en) | Copper wire tinning defect identification method | |
CN115841633A (en) | Power tower and power line associated correction power tower and power line detection method | |
CN116993731A (en) | Shield tunneling machine tool bit defect detection method based on image | |
CN111242891A (en) | Method for identifying and classifying surface defects of steel rails | |
CN117218029B (en) | Night dim light image intelligent processing method based on neural network | |
CN116309447B (en) | Dam slope crack detection method based on deep learning | |
CN116485884B (en) | Real-time positioning method and system for finish brewing beer bottle mouth based on computer vision |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |