CN108596221B - Image recognition method and device for scale reading - Google Patents
Image recognition method and device for scale reading Download PDFInfo
- Publication number
- CN108596221B CN108596221B CN201810315600.5A CN201810315600A CN108596221B CN 108596221 B CN108596221 B CN 108596221B CN 201810315600 A CN201810315600 A CN 201810315600A CN 108596221 B CN108596221 B CN 108596221B
- Authority
- CN
- China
- Prior art keywords
- scale
- image
- length
- measured
- reading
- 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
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000001514 detection method Methods 0.000 claims abstract description 45
- 238000005259 measurement Methods 0.000 claims abstract description 22
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 24
- 238000004422 calculation algorithm Methods 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 15
- 238000003860 storage Methods 0.000 claims description 8
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000008450 motivation Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an image identification method and device for scale reading. The method comprises the following steps: acquiring an image of an object to be measured by using a scale; identifying the length of a scale image in the image of the object to be measured along the scale measuring direction by using a scale image detection model, and intercepting the scale image from the image of the object to be measured; identifying the length of a scale image in the intercepted scale image along the scale measuring direction by using a scale image detection model; and calculating to obtain the reading of the object to be measured by the scale according to the length of the scale image along the scale measuring direction, the actual total length of the scale and the actual length of the scale in the scale. The measurement reading of the scale is calculated by identifying the length of the scale and the length of the scale, so that the adaptability of scale image identification to a complex environment can be improved.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to an image recognition method and device for scale reading.
Background
With the development of the information age, image data is becoming increasingly voluminous and growing at a rapid rate. The thinking of people can be found out through the images, and a basis is provided for further decision making. Before data mining, effective classification management of such enormous image data is required. In the classification management of image data, image recognition is a basic work, and has important significance and value in information extraction, information recognition and information retrieval. It is difficult for a computer to perform image classification or image recognition on image data like a person in that the image data has irregularities, representation methods of different image data, orders of magnitude of images are different, and the like. Deep Learning (Deep Learning), especially Convolutional Neural Network (CNN), is a research focus in image processing and pattern recognition in recent years, and is receiving more and more attention. The neural network is a neural network which is established and simulates the human brain to analyze and learn and can simulate the mechanism of the human brain to interpret data such as images, sounds and texts.
The scale image intelligent recognition system is a scale image automatic recognition product facing the water industry and the energy industry in a plurality of image recognition products, can automatically recognize the reading in a water scale image, and carries out professional business application on the recognized result, can replace the traditional manual observation counting, improves the working efficiency of users, and reduces manual operation.
From the hierarchical structure of the machine learning model, the development of machine learning has undergone probably two changes: shallow learning and deep learning. When the heterogeneous problem of a plurality of variables is faced, the shallow learning model is difficult to express, a network with a plurality of layers of hidden nodes is needed to effectively express the variables, and the deep learning model can simply and effectively express complex functions. With the development of big data, in the face of huge data, a shallow learning structure shows a short board in terms of model description capability, internal representation of the data is difficult to fully mine, and more valuable information can be mined from the big data only by a model with stronger expression capability, so that a learning motivation of exploring a deep learning model to model a complex function is stimulated. The research goal of image recognition is to classify images into different predefined categories according to some kind of attribute they have. How to construct image feature representation and classification models is the key to solving the image understanding problem, and many researchers have conducted extensive research and proposed some effective methods. Most of the traditional methods are based on visual codebook models, the models well utilize image description and effective machine learning models which are artificially and skillfully designed, but the expression force of the models on the middle-layer structure and high-layer semantic information of the images is limited, and the semantic gap can not be broken through. In recent years, the breakthrough development of deep learning provides a new idea for solving the problem, and the method is successfully applied to a plurality of pattern recognition problems.
However, in the existing image recognition product based on deep learning, the strong assumed space of the network model makes the model training easily fall into local optimization, so that the generalization capability of the model is poor, and unknown data cannot be well predicted. Meanwhile, a large amount of data containing semantic labels are needed for training model training, and the workload of manual labeling is increased. In the existing scale recognition method based on hough transformation and harris detection, after median filtering is used for removing noise and gray level equalization, morphological thinning and contour extraction technologies are used for calculating the position of a scribed line. The algorithm adopts a traditional image processing method, is good in performance only on a small number of images, and cannot be well adapted to changes of various water gauge models (such as water gauge color, scales, sizes and the like) and complex environments (such as illumination, angles and the like). The existing scale positioning algorithm based on scale colors of a scale adopts a color segmentation method to extract color channel characteristics in an image and perform threshold segmentation to position the scale, but the algorithm cannot be well adapted to a complex environment, for example, when stains, spots, local damage or various scale colors exist on the scale, the robustness of the algorithm is not high.
Disclosure of Invention
The invention provides a method and equipment for identifying an image of a scale reading, which are used for improving the adaptability of scale image identification to a complex environment.
The embodiment of the invention provides an image identification method of scale reading, which comprises the following steps: acquiring an image of an object to be measured by using a scale; identifying the length of a scale image in the image of the object to be measured along the scale measuring direction by using a scale image detection model, and intercepting the scale image from the image of the object to be measured; identifying the length of a scale image in the intercepted scale image along the scale measuring direction by using a scale image detection model; and calculating to obtain the reading of the object to be measured by the scale according to the length of the scale image along the scale measuring direction, the actual total length of the scale and the actual length of the scale in the scale.
In one embodiment, the method further comprises: training an initial neural network by using first training data and an image target detection algorithm to obtain the scale image detection model; the first training data includes a scale measurement image and a corresponding scale frame label.
In one embodiment, the method further comprises: training an initial neural network by using second training data and an image target detection algorithm to obtain the scale image detection model; the second training data includes scale images in the scale measurement image and corresponding scale box labels.
In one embodiment, identifying a length of a scale image in a truncated scale image along a scale measurement direction using a scale image detection model comprises: identifying scale frame coordinates of at least one of the intercepted scale images using a scale image detection model; calculating the initial length of each scale image along the scale measuring direction according to the coordinates of each scale frame; and obtaining the length of the scale image in the intercepted scale image along the scale measuring direction according to the initial length corresponding to at least one scale image.
In one embodiment, the initial neural network is a convolutional neural network; the image target detection algorithm is a fast RCNN target detection algorithm; the method further comprises the following steps: and obtaining the initial neural network by using a ZFNET algorithm based on a set network structure.
In one embodiment, the scale measures the readings of the object to be measured as:
Hw＝H–Hr/He*scale，
wherein Hw represents the reading of the object to be measured by the scale, H represents the actual total length of the scale, Hr represents the length of the scale image along the scale measuring direction, He represents the length of the scale image along the scale measuring direction, and scale represents the actual length of the scale in the scale.
In one embodiment, the method further comprises: and when the reading exceeds a set value, sending early warning information to the terminal equipment.
In one embodiment, acquiring an image of an object to be measured using a scale comprises: acquiring a video image of an object to be measured by using a scale; and converting the video image into an image of the object to be measured.
Embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the method described in the above embodiments.
The embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and when the processor executes the computer program, the steps of the method described in the above embodiments are implemented.
According to the image identification method of the scale reading, the computer readable storage medium and the computer device, the measurement reading of the scale is calculated by identifying the length of the scale and the length of the scale, the method is not affected by the change of light on the surface of an object to be measured, the measurement angle and the like, and the adaptability of the scale image identification to a complex environment is improved.
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 description of the embodiments or 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. In the drawings:
fig. 1 is a flowchart illustrating an image recognition method for reading a scale according to an embodiment of the present invention.
Fig. 2 is a flowchart illustrating a method for identifying a length of a scale image in a truncated scale image along a scale measurement direction by using a scale image detection model according to an embodiment of the present invention.
Fig. 3 is a flowchart illustrating an image recognition method for reading a scale according to another embodiment of the present invention.
Fig. 4 is a flowchart illustrating a method for obtaining an image of an object to be measured by using a ruler according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. The exemplary embodiments and descriptions of the present invention are provided to explain the present invention, but not to limit the present invention.
Fig. 1 is a flowchart illustrating an image recognition method for reading a scale according to an embodiment of the present invention. As shown in fig. 1, the method for image recognition of scale reading may include:
step S110: acquiring an image of an object to be measured by using a scale;
step S120: identifying the length of a scale image in the image of the object to be measured along the scale measuring direction by using a scale image detection model, and intercepting the scale image from the image of the object to be measured;
step S130: identifying the length of a scale image in the intercepted scale image along the scale measuring direction by using a scale image detection model;
step S140: and calculating to obtain the reading of the object to be measured by the scale according to the length of the scale image along the scale measuring direction, the actual total length of the scale and the actual length of the scale in the scale.
The scale is, for example, a water gauge, and the object to be measured is, for example, a reservoir, a river, or the like. The scale may refer to a unit of measure on a scale, such as the scale E on a water gauge. When the scale is used for measuring an object to be measured, only a part of the scale is generally visible, for example, when the water gauge is used for measuring the depth of a reservoir, only the water gauge above the water surface is generally visible, so that the scale image in the image of the object to be measured generally only contains the visible part of the scale. The length of the scale image in the scale measuring direction and the length of the scale image in the scale measuring direction may be determined according to the distance between two points in the image. The actual total length of the scale and the actual length of the scale in the scale may be determined from the actual scale. The image of the object to be measured may include one or more scale images. The scale portion of the scale image should contain one or more sharp scales, and there may be one or more scale imperfections or deletions in the scale portion of the scale image. The scale image detection model and the scale image detection model may be parametric models for a certain network structure.
In the embodiment, by using the scale image detection model to identify the length of the scale image along the scale measuring direction in the image of the object to be measured, intercepting the scale image from the image of the object to be measured, identifying the length of the scale image in the intercepted scale image along the measuring direction of the scale by using a scale image detection model, and the reading of the object to be measured by the scale is calculated according to the length of the scale image along the scale measuring direction, the actual total length of the scale and the actual length of the scale in the scale, so that the measurement reading of the scale is calculated by identifying the length of the scale and the length of the scale, the influence of surface light change, measurement angle and the like of the object to be measured is avoided, and the adaptability of scale image identification to a complex environment is improved.
In an embodiment, in the above steps S120 and S130, the confidence threshold conf _ thresh during prediction may be 0.7 to 0.9, for example, 0.80, and the non-maximum suppression threshold nms _ thresh may be 0.2 to 0.4, for example, 0.30.
In some embodiments, the method for image recognition of scale readings may further comprise: training an initial neural network by using first training data and an image target detection algorithm to obtain the scale image detection model; the first training data includes a scale measurement image and a corresponding scale frame label. The scale measurement image is an image of an object to be measured using a scale. Every scale frame label can mark the position of scale in the scale measurement image, can come out the scale frame.
In some embodiments, the method for image recognition of scale readings may further comprise: training an initial neural network by using second training data and an image target detection algorithm to obtain the scale image detection model; the second training data includes scale images in the scale measurement image and corresponding scale box labels. Each scale box label may indicate a position or box coordinate of the scale image.
The second training data may be obtained based on the first training data. The initial neural network used to obtain the scale image detection model and the scale image detection model may be the same. In training, the number of iterations may be set to 5 ten thousand, for example.
Fig. 2 is a flowchart illustrating a method for identifying a length of a scale image in a truncated scale image along a scale measurement direction by using a scale image detection model according to an embodiment of the present invention. As shown in fig. 2, in the step S130, identifying the length of the scale image in the scale measuring direction in the cut scale image by using the scale image detection model may include:
step S131: identifying frame coordinates of at least one scale image in the intercepted scale images by using a scale image detection model;
step S132: calculating the initial length of each scale image along the scale measuring direction according to the coordinates of each scale frame;
step S133: and obtaining the length of the scale image in the intercepted scale image along the scale measuring direction according to the initial length corresponding to at least one scale image.
The scale portion of the truncated scale image may contain one or more sharp scales. The scale box coordinates may include coordinates of points on two or more scale boxes. The initial length of the scale image along the scale measurement direction may be the difference between the ordinate of two points in the scale frame coordinates. In the above step S133, for example, the length of the scale image in the scale measuring direction may be obtained by averaging the initial lengths corresponding to a plurality of the scale images; or one of the initial lengths is selected as the length of the scale image along the measuring direction of the scale.
In some embodiments, the initial neural network is a convolutional neural network. In some embodiments, the image object detection algorithm is a fast RCNN object detection algorithm.
In some embodiments, the method for image recognition of scale readings may further comprise: and obtaining the initial neural network by using a ZFNET algorithm based on a set network structure.
In some embodiments, the scale measures readings of the object to be measured as:
Hw＝H–Hr/He*scale，
wherein Hw represents the reading of the object to be measured by the scale, H represents the actual total length of the scale, Hr represents the length of the scale image along the scale measuring direction, He represents the length of the scale image along the scale measuring direction, and scale represents the actual length of the scale in the scale.
In this embodiment, for example, when the water gauge is used to measure the depth of the water body, the total length of the water gauge minus the length of the part of the water gauge (Hr/He scale) above the water surface can be used to obtain the reading of the water gauge measuring the water body, i.e. the water depth.
Fig. 3 is a flowchart illustrating an image recognition method for reading a scale according to another embodiment of the present invention. As shown in fig. 3, the image recognition method for reading the scale shown in fig. 1 may further include:
step S150: and when the reading exceeds a set value, sending early warning information to the terminal equipment.
The terminal device may be a mobile terminal device, such as a mobile phone. Further, the early warning information may be sent to an APP (application) on the terminal device, such as a short message, a WeChat, and the like. Therefore, when the threat of the object to be detected to people is known through the scale measurement reading, relevant personnel can be prompted to take corresponding measures in time.
Fig. 4 is a flowchart illustrating a method for obtaining an image of an object to be measured by using a ruler according to an embodiment of the present invention. As shown in fig. 4, in the step S110, acquiring an image of the object to be measured by using the ruler may include:
step S111: acquiring a video image of an object to be measured by using a scale;
step S112: and converting the video image into an image of the object to be measured.
In this embodiment, the video image may be obtained from a monitoring point of the object to be measured. The image of the object to be measured may be obtained by capturing a picture from a video image. And capturing the images of the plurality of objects to be measured at different playing time points.
In one embodiment, taking a water gauge as an example, the deep learning-based automatic water gauge image recognition algorithm may include:
step S1: defining a water gauge image recognition network structure, and using ZFNET as a basic training network in an algorithm;
step S2: respectively training the scale and the scale E by using a faster rcnn target detection algorithm according to the network structure and the parameters configured in the step S1, wherein the iteration times are set to 50000 times, and thus a scale and scale E detection model is obtained;
step S3: sending the scale detection model obtained in the step S2 and the input image I into a network for forward calculation, and obtaining the regression frame coordinates of the scale and intercepting the scale image, wherein a confidence coefficient threshold conf _ thresh is 0.80 and a non-maximum value inhibition threshold nms _ thresh is 0.30 during prediction;
step S4: sending the scale E detection model obtained in the step S2 and the scale image intercepted in the step S3 into a convolutional neural network for forward calculation, obtaining a confidence coefficient threshold conf _ thresh of 0.80 and a non-maximum suppression threshold nms _ thresh of 0.30 during prediction, obtaining regression frame coordinates BBoxi (x1, y1, x2 and y2) of N scales E on the scale, and calculating the average height calculation formula of the scale E frame as He-1/N sigma (y 2-y 1);
step S5: and according to the step S3 and the step S4, the position of the scale, the frame height Hr of the scale in the scale image and the average height He of the frame of the scale E are obtained through forward calculation, the unit length of the scale E is given as scale, the total height of the water gauge is given as H, and the water depth calculation formula is as follows: Hw-Hr/He scale (unit m).
According to the method, the computer image automatic identification technology is utilized to automatically identify the readings of various water scales in the water industry, so that the traditional manual naked eye judgment of the readings is replaced, the early warning condition is set, the early warning is automatically analyzed, the labor energy investment for observing the water scales can be gradually reduced, and the working efficiency is really improved. Can promote water gauge video monitoring ability and dynamics by a wide margin, the discovery is unusual can be more timely, accurate. The function of the water gauge video monitoring station can be exerted to the greatest extent.
In some embodiments, the scale image intelligent recognition system based on the scale reading image recognition method mainly comprises a video image access part, an image recognition calculation model, a calculation result analysis early warning part and a calculation result display part. The access of the video image can mean that the intelligent ruler image recognition system is connected with the video monitoring point equipment to acquire video image information in real time, and the system automatically converts the video image into pictures for picture content recognition. The image recognition calculation model can be an automatic recognition calculation model established aiming at the water scale image, so that a computer can automatically recognize the reading of the water scale in the picture, and the computer can continuously learn and optimize by self to improve the recognition accuracy. The calculation result analysis and early warning can be to automatically judge whether the threshold value is exceeded or not by setting an early warning threshold value aiming at the recognized scale reading, automatically generate early warning information after the threshold value is exceeded, and send related personnel through WeChat and short message. The calculation result display can be directed at the achievement information of automatic recognition calculation of the system, and intuitive display and query are carried out through the web.
Embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the method described in the above embodiments.
The embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and when the processor executes the computer program, the steps of the method described in the above embodiments are implemented.
In summary, the image recognition method for reading a scale, the computer-readable storage medium and the computer device according to the embodiments of the present invention calculate the measurement reading of the scale by recognizing the length of the scale and the length of the scale, so that the method is not affected by the change of light on the surface of the object to be measured, the measurement angle, and the like, and the adaptability of the scale image recognition to a complex environment is improved.
In the description herein, reference to the description of the terms "one embodiment," "a particular embodiment," "some embodiments," "for example," "an example," "a particular example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. The sequence of steps involved in the various embodiments is provided to schematically illustrate the practice of the invention, and the sequence of steps is not limited and can be suitably adjusted as desired.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (7)
1. An image recognition method for scale reading, comprising:
acquiring an image of an object to be measured by using a scale, wherein the scale is a water gauge;
identifying the length of a scale image in the image of the object to be measured along the scale measurement direction by using a scale image detection model, and intercepting the scale image from the image of the object to be measured, wherein an initial neural network is trained by using first training data and an image target detection algorithm to obtain the scale image detection model, and the first training data comprises a scale measurement image and a corresponding scale frame label;
identifying the length of a scale image in the intercepted scale image along the scale measuring direction by using a scale image detection model, wherein the initial neural network is trained by using second training data and an image target detection algorithm to obtain the scale image detection model, and the second training data comprises the scale image in the scale measurement image and a corresponding scale frame label;
calculating to obtain the reading of the object to be measured by the scale according to the length of the scale image along the scale measuring direction, the actual total length of the scale and the actual length of the scale in the scale;
identifying a length of a scale image in the intercepted scale image along a scale measurement direction using a scale image detection model, comprising:
identifying scale frame coordinates of at least one of the intercepted scale images using a scale image detection model;
calculating the initial length of each scale image along the scale measuring direction according to the coordinates of each scale frame;
and obtaining the length of the scale image in the intercepted scale image along the scale measuring direction according to the initial length corresponding to at least one scale image.
2. A method of image recognition of a scale reading according to claim 1, wherein the initial neural network is a convolutional neural network; the image target detection algorithm is a fast RCNN target detection algorithm;
the method further comprises the following steps:
and obtaining the initial neural network by using a ZFNET algorithm based on a set network structure.
3. The method for image recognition of a reading of a scale according to claim 1, wherein the scale measuring the reading of the object to be measured is:
Hw＝H-Hr/He*scale，
wherein Hw represents the reading of the object to be measured by the scale, H represents the actual total length of the scale, Hr represents the length of the scale image along the scale measuring direction, He represents the length of the scale image along the scale measuring direction, and scale represents the actual length of the scale in the scale.
4. A method of image recognition of a scale reading according to claim 1, further comprising:
and when the reading exceeds a set value, sending early warning information to the terminal equipment.
5. The method of image recognition of a scale reading according to claim 1, wherein obtaining an image of an object to be measured using the scale comprises:
acquiring a video image of an object to be measured by using a scale;
and converting the video image into an image of the object to be measured.
6. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method of claims 1 to 5.
7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of claims 1 to 5 are implemented when the processor executes the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810315600.5A CN108596221B (en) | 2018-04-10 | 2018-04-10 | Image recognition method and device for scale reading |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810315600.5A CN108596221B (en) | 2018-04-10 | 2018-04-10 | Image recognition method and device for scale reading |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108596221A CN108596221A (en) | 2018-09-28 |
| CN108596221B true CN108596221B (en) | 2020-12-01 |
Family
ID=63621470
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810315600.5A Active CN108596221B (en) | 2018-04-10 | 2018-04-10 | Image recognition method and device for scale reading |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108596221B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109214806B (en) * | 2018-11-20 | 2022-01-04 | 北京京东尚科信息技术有限公司 | Self-service settlement method, device and storage medium |
| CN109784259B (en) * | 2019-01-08 | 2021-04-13 | 江河瑞通（北京）技术有限公司 | Intelligent water transparency identification method based on image identification and Samsung disk assembly |
| CN109859183B (en) * | 2019-01-29 | 2021-06-04 | 江河瑞通（北京）技术有限公司 | Edge calculation-based multi-element integrated water body intelligent identification method and ecological station |
| CN110223341B (en) * | 2019-06-14 | 2024-05-28 | 北京国信华源科技有限公司 | Intelligent water level monitoring method based on image recognition |
| CN111877105A (en) * | 2020-06-16 | 2020-11-03 | 中交第三公路工程局有限公司 | Modified asphalt construction method |
| CN113819972A (en) * | 2020-07-07 | 2021-12-21 | 湖北亿立能科技股份有限公司 | Artificial intelligence surface of water detecting system based on virtual scale |
| CN113112514B (en) * | 2021-04-27 | 2024-05-17 | 汇鸿智能科技(辽宁)有限公司 | Method, device, computer equipment and storage medium for identifying graphite size by AI |
| CN113776408B (en) * | 2021-09-13 | 2022-09-13 | 北京邮电大学 | Reading method for gate opening ruler |
| CN115240394B (en) * | 2022-09-22 | 2022-12-20 | 国网湖北省电力有限公司经济技术研究院 | Method and system for monitoring and early warning water level of accident oil pool of transformer substation |
| CN116469091A (en) * | 2023-04-21 | 2023-07-21 | 浪潮智慧科技有限公司 | Automatic water gauge reading method, device and medium based on real-time video |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105741316A (en) * | 2016-01-20 | 2016-07-06 | 西北工业大学 | Robust target tracking method based on deep learning and multi-scale correlation filtering |
| CN105938554A (en) * | 2016-04-14 | 2016-09-14 | 北京交通大学 | Switch rail expansion displacement monitoring method based on image automatic interpretation and switch rail expansion displacement monitoring system thereof |
| CN105973342A (en) * | 2016-06-03 | 2016-09-28 | 湖南相水缘水利科技有限公司 | Intelligent water level monitoring system |
| CN106295566A (en) * | 2016-08-10 | 2017-01-04 | 北京小米移动软件有限公司 | Facial expression recognizing method and device |
| CN106599892A (en) * | 2016-12-14 | 2017-04-26 | 四川长虹电器股份有限公司 | Television station logo identification system based on deep learning |
| CN107038751A (en) * | 2015-12-07 | 2017-08-11 | 达索系统公司 | The identification of 3D modeling object is carried out from 2D images |
| CN107122806A (en) * | 2017-05-16 | 2017-09-01 | 北京京东尚科信息技术有限公司 | A kind of nude picture detection method and device |
| CN107367310A (en) * | 2017-07-11 | 2017-11-21 | 华南理工大学 | A kind of river level remote monitoring method based on computer vision |
| CN107506798A (en) * | 2017-08-31 | 2017-12-22 | 福建四创软件有限公司 | A kind of water level monitoring method based on image recognition |
| CN107607207A (en) * | 2017-08-22 | 2018-01-19 | 中国科学院深圳先进技术研究院 | A kind of power equipment thermal fault detection method, system and electronic equipment |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4054008B2 (en) * | 2004-08-06 | 2008-02-27 | 株式会社サンキキ | Water level monitoring device using a quantitative water mark |
| CN100573183C (en) * | 2006-03-09 | 2009-12-23 | 华北电力大学 | Full-automatic calibration method of direct-reading instrument and device |
-
2018
- 2018-04-10 CN CN201810315600.5A patent/CN108596221B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107038751A (en) * | 2015-12-07 | 2017-08-11 | 达索系统公司 | The identification of 3D modeling object is carried out from 2D images |
| CN105741316A (en) * | 2016-01-20 | 2016-07-06 | 西北工业大学 | Robust target tracking method based on deep learning and multi-scale correlation filtering |
| CN105938554A (en) * | 2016-04-14 | 2016-09-14 | 北京交通大学 | Switch rail expansion displacement monitoring method based on image automatic interpretation and switch rail expansion displacement monitoring system thereof |
| CN105973342A (en) * | 2016-06-03 | 2016-09-28 | 湖南相水缘水利科技有限公司 | Intelligent water level monitoring system |
| CN106295566A (en) * | 2016-08-10 | 2017-01-04 | 北京小米移动软件有限公司 | Facial expression recognizing method and device |
| CN106599892A (en) * | 2016-12-14 | 2017-04-26 | 四川长虹电器股份有限公司 | Television station logo identification system based on deep learning |
| CN107122806A (en) * | 2017-05-16 | 2017-09-01 | 北京京东尚科信息技术有限公司 | A kind of nude picture detection method and device |
| CN107367310A (en) * | 2017-07-11 | 2017-11-21 | 华南理工大学 | A kind of river level remote monitoring method based on computer vision |
| CN107607207A (en) * | 2017-08-22 | 2018-01-19 | 中国科学院深圳先进技术研究院 | A kind of power equipment thermal fault detection method, system and electronic equipment |
| CN107506798A (en) * | 2017-08-31 | 2017-12-22 | 福建四创软件有限公司 | A kind of water level monitoring method based on image recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108596221A (en) | 2018-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108596221B (en) | Image recognition method and device for scale reading | |
| CN113269073B (en) | Ship multi-target tracking method based on YOLO V5 algorithm | |
| CN111160440B (en) | Deep learning-based safety helmet wearing detection method and device | |
| CN105975913B (en) | Road network extraction method based on adaptive cluster learning | |
| CN110458095A (en) | A kind of recognition methods, control method, device and the electronic equipment of effective gesture | |
| CN110569843B (en) | Intelligent detection and identification method for mine target | |
| CN101986348A (en) | Visual target identification and tracking method | |
| CN103105924B (en) | Man-machine interaction method and device | |
| CN112149512A (en) | Helmet wearing identification method based on two-stage deep learning | |
| CN114998852A (en) | Intelligent detection method for road pavement diseases based on deep learning | |
| KR102346676B1 (en) | Method for creating damage figure using the deep learning-based damage image classification of facility | |
| CN110610165A (en) | Ship behavior analysis method based on YOLO model | |
| CN115600513B (en) | Karst collapse monitoring early warning and prevention and control integrated informationized simulation research and judgment system | |
| CN104038792A (en) | Video content analysis method and device for IPTV (Internet Protocol Television) supervision | |
| CN110458022A (en) | It is a kind of based on domain adapt to can autonomous learning object detection method | |
| CN111931573A (en) | Helmet detection and early warning method based on YOLO evolution deep learning model | |
| CN102968786B (en) | A kind of non-supervisory remote sensing images potential target method for detecting area | |
| CN113160220A (en) | Door handle homing and bending detection method based on deep learning | |
| CN115841557B (en) | Intelligent crane operation environment construction method based on digital twin technology | |
| CN117475353A (en) | Video-based abnormal smoke identification method and system | |
| CN113887634B (en) | Electric safety belt detection and early warning method based on improved two-step detection | |
| CN115690514A (en) | Image recognition method and related equipment | |
| Hake et al. | Damage detection for port infrastructure by means of machine-learning-algorithms | |
| CN113920450A (en) | Method and device for identifying insulator RTV coating based on intrinsic image decomposition | |
| CN116152674A (en) | Dam unmanned aerial vehicle image crack intelligent recognition method based on improved U-Net model |
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 | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 100097 room 10e, unit 2 (block B), 9 / F, building 2, yard 2, lantianchang East Road, Haidian District, Beijing (Building 2, Jinyuan times business center) Patentee after: Kunyu (Beijing) Technology Co.,Ltd. Address before: A646, 7th Floor, 7th Floor, No. 28 Xinxin Road, Haidian District, Beijing, 100085 Patentee before: RICHWAY (BEIJING) TECHNOLOGY Co.,Ltd. |