CN112329787A - Water gauge character positioning method based on information attention - Google Patents

Water gauge character positioning method based on information attention Download PDF

Info

Publication number
CN112329787A
CN112329787A CN202011413570.5A CN202011413570A CN112329787A CN 112329787 A CN112329787 A CN 112329787A CN 202011413570 A CN202011413570 A CN 202011413570A CN 112329787 A CN112329787 A CN 112329787A
Authority
CN
China
Prior art keywords
information
character
attention
map
water level
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.)
Granted
Application number
CN202011413570.5A
Other languages
Chinese (zh)
Other versions
CN112329787B (en
Inventor
刘阳
郑全新
张磊
宁志勇
董小栋
张逞逞
牛小芳
孟祥松
朱浩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Tongfang Software Co Ltd
Original Assignee
Beijing Tongfang Software Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Tongfang Software Co Ltd filed Critical Beijing Tongfang Software Co Ltd
Priority to CN202011413570.5A priority Critical patent/CN112329787B/en
Publication of CN112329787A publication Critical patent/CN112329787A/en
Application granted granted Critical
Publication of CN112329787B publication Critical patent/CN112329787B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • G06V30/14—Image acquisition
    • G06V30/148—Segmentation of character regions
    • G06V30/153—Segmentation of character regions using recognition of characters or words
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Character Discrimination (AREA)

Abstract

A water level ruler character positioning method based on the information attention degree relates to the field of artificial intelligence and computer vision. The method comprises the following steps: the camera captures images of the water level gauge; secondly, overlapping the upper layer map information and the lower layer map information of the obtained position characteristic map, and sending the information to an upper information attention module for character positioning; thirdly, carrying out differential attention on the position information captured by the local layer of features and all the upper layer of features in the position feature map to obtain accurate character position information; and (IV) carrying out character recognition on the scratched image, and acquiring the scale of the scale to finish output. The invention adds the above information attention module on the basis of the depth detector, and the module carries out difference attention on the position information characteristic map and accurately positions the area where the water level gauge character is positioned; meanwhile, the information attention module comprehensively judges the position information, so that the influence of the change of environmental factors such as illumination on the detection performance is greatly weakened.

Description

Water gauge character positioning method based on information attention
Technical Field
The invention relates to the field of artificial intelligence and the field of computer vision, in particular to a method for positioning water level ruler characters through the information attention degree.
Background
The invention provides a water level identification method based on image identification, which is invented in 'a water level identification method based on image identification' applied by Nanjing forestry university in 2019, and comprises the following steps: positioning the water gauge position from the water gauge image, and cutting the water gauge image according to the positioned water gauge position; character positioning is carried out on the water gauge image obtained by cutting, each character in the water gauge image is obtained, and each digital character in the water gauge image is cut according to the arrangement characteristics of the characters, so that each digital character image is obtained; identifying the digital characters in each digital character image to obtain the numerical values of each digital character; and identifying the water mark position in the water gauge image, and calculating the water level height according to the relative position relation between the water mark position and the lowermost character. The water level identification method based on image identification combines MSER and a template matching algorithm, can meet the requirement of water gauge positioning in a complex scene, well solves the problem of character identification errors caused by light reflection and stains of individual characters, and has good robustness.
The invention discloses a water level identification monitoring system, which is invented patent 'a water level identification monitoring system' applied by Wuhan century hydronic science and technology Limited company in 2019 and comprises a video image data acquisition layer, a data transmission processing layer and a monitoring data display layer, wherein the output end of the video image data acquisition layer is connected with the input end of the data transmission processing layer through an internet or a USB interface, and the output end of the data transmission processing layer is connected with the input end of the monitoring data display layer through the internet. This water level discernment monitoring system can improve the water level monitoring precision greatly, need not carry out signal conversion, data acquisition and transmission are more reliable and more stable, can discern according to high-precision camera image, the backstage program has the artificial intelligence algorithm of adoption constantly to go deep into study, can self-adjustment in order to adapt to complex environment, adaptability to the environment is higher, this water level monitoring system installation only need install high-precision camera and installation software simultaneously can, need not carry out the construction of extra monitoring well, it is convenient to install, the construction is simple.
The invention provides a water level identification method and a water level identification system based on a virtual water gauge, which are invented in 2019 by Jiangheliantong (Beijing) technology Limited company, and the method comprises the following steps: selecting two initial points of known actual positions in a natural water area and image coordinates of the two initial points in a water area image corresponding to the natural water area; identifying a water area region in the water area image through a water area segmentation model; according to the water area, the actual positions of the two initial points and the image coordinates, the water surface elevation of the current water surface of the natural water area is determined.
The invention discloses a hydrological ruler water level identification and estimation method based on image analysis, which is invented in ' a hydrological ruler water level identification and estimation method based on image analysis ' applied by national defense science and technology university of China people's liberation force in 2018, and comprises the following steps: firstly, realizing initial positioning of a hydrological ruler through an HSV color space, and then adaptively constructing a component diagram to realize accurate positioning of the hydrological ruler; expanding samples by a data enhancement technology, realizing character segmentation on the positioned hydrological ruler image by adopting a fuzzy C clustering method, then constructing a convolutional neural network training data set and identifying complete characters; and sequentially calculating the pixel heights of the complete characters E, establishing a mapping relation between the pixel heights and the actual heights in the quadratic equation dynamic fitting image on the basis of the pixel heights of the complete characters E, and further calculating the actual heights of the incomplete characters E. The invention can realize the recognition and estimation of the water depth in the image containing the hydrological gauge, and is suitable for being integrated at the front end of a terminal, such as a video camera, a camera and the like, to automatically estimate the water level of the hydrological gauge.
The invention discloses a water level identification method based on a water gauge image, which is disclosed in 'water level identification method based on a water gauge image' filed by south Water science and technology Limited in Jiangsu in 2018, and comprises the following steps: carrying out binarization processing; performing morphological analysis; cutting out; measuring and calculating the slope; correcting the position; accurately positioning; image cutting; recognizing nerves; and a water level determining step. The water level identification method provided by the invention realizes the non-contact automatic identification of the water level in the field, has accurate identification result, reduces the cost and simplifies the hardware structure; the method of the invention can adapt to automatic water level monitoring under various conditions, and promotes the development of water level monitoring technology.
The prior art described above also has the following drawbacks:
1. the current water level gauge scale positioning method mostly adopts image processing to obtain required various character positions, and a few methods use a detector to position characters on a gauge, but the environment where the water level gauge is located is basically a bridge, a culvert, a tunnel portal and other places with large traffic flow, along with the change of the illumination intensity of a car lamp, the illumination environment where the whole water level gauge is located is extremely unstable, and the characters cannot be positioned easily.
2. The camera mounting height is far above the scale, shoots the scale with very big angle of depression basically, and the character on the scale will demonstrate very big distortion on shooting the picture, and the size of the character on the scale on the image is less than the area of whole water level chi far away to there is fuzzification to some extent, and above circumstances is a very big challenge to character accurate positioning, will directly influence final water level precision.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention aims to provide a water level ruler character positioning method based on the information attention degree. The depth detector is added with the information attention module on the basis of the depth detector, and the module carries out difference attention on the position information characteristic map and accurately positions the area where the water level gauge characters are located; meanwhile, the information attention module comprehensively judges the position information, so that the influence of the change of environmental factors such as illumination on the detection performance is greatly weakened.
In order to achieve the above object, the technical solution of the present invention is implemented as follows:
a water level ruler character positioning method based on the above information attention degree uses a position information extraction module, an above information attention degree module and a water level ruler scale identification module, and the specific method comprises the following steps:
the camera captures images of the water level gauge;
and (II) extracting the position characteristic information of the image on different scales by a position information extraction module to obtain a position characteristic map. Superposing upper layer map information and lower layer map information in the position characteristic map and sending the superposed upper layer map information and lower layer map information to an upper information attention module for character positioning;
and thirdly, the information attention module carries out differential attention on the position information captured by the local layer of features and all the upper layer of features in the position feature map to obtain accurate character position information. The above information attention module comprises the following processing steps:
1) to the atlas F in the position characteristic atlas1All characteristic maps F of upper layer2,F3,…FnPerforming upsampling to obtain an upsampled value1Feature map Fu with same dimensions2，Fu3…Fun。
2) F is to be1，Fu2,Fu3…FunAnd performing element superposition to obtain a characteristic Fm. Performing convolution operation on Fm to obtain Fm ', Fm' scale and F1The size of the scale is the same, the number of channels is n, n is equal to the input F of the module1,F2,F3,…FnThe sum of the number of the characteristic spectrum channels is the same.
3) And (3) carrying out normalization processing on Fm' to obtain a difference attention degree score of the position characteristic map, wherein the score represents the utilization rate of the position information contained in the position characteristic map.
4) The normalized score is compared to the feature profile F1，Fu2,Fu3…FunAnd (4) carrying out element multiplication, and extracting the difference attention feature Fdot.
5) And superposing the characteristics Fdot to obtain character positioning information.
And (IV) using the character positioning information to carry out character scratching on the image, sending the scratched image into a water level scale identification module to carry out character identification, and acquiring scale scales to finish output.
In the water level ruler character positioning method based on the above information attention, the above information attention module performs differential attention on the local layer of features and all the features on the upper layer in the position feature map to obtain accurate character position information.
Due to the adoption of the method, compared with the prior art, the method has the following advantages:
1. the invention completes the difference attention to the position information through the information attention module, and realizes the accurate positioning of the character position.
2. According to the invention, the information attention module comprehensively judges the position information, so that the influence of factors such as environmental illumination and distortion on positioning is effectively eliminated, and the positioning performance is more robust.
The invention is further described with reference to the following figures and detailed description.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic processing diagram of a location information extraction module according to an embodiment of the present invention;
FIG. 3 is a block diagram illustrating an exemplary structure of the above information attention module according to an embodiment of the present invention;
fig. 4 is a diagram of a depth network structure of a water level scale identification module in the embodiment of the invention.
Detailed Description
Referring to fig. 1, the water level gauge character positioning method based on the above information attention degree of the present invention uses a position information extraction module, an above information attention degree module and a water level gauge scale identification module, and the specific method steps are as follows:
the camera captures images of the water level gauge.
And (II) transmitting the captured image into a position information extraction module to extract a position characteristic map.
(III) the above information attention module carries out difference attention on the position information captured by the local layer of features and all the upper layer of features in the position feature map to obtain accurate character position information
And (IV) using the character positioning information to carry out character scratching on the image, sending the scratched image into a character recognition module to carry out character recognition, and acquiring scale scales of the scale to complete output.
The structure and principle of the three aspects of the position information extraction module, the information attention module and the water level scale identification module are described in detail in the following.
Position information extraction module
Referring to fig. 2, the module mainly extracts the position information by using a feature pyramid structure. The object dimension processed from bottom to top of the feature pyramid structure is gradually increased, namely the bottom layer of the feature pyramid can effectively capture the position information of the object with smaller dimension, and the top layer of the pyramid can effectively capture the position information of the object with larger dimension. Therefore, the characteristic pyramid can effectively extract the position information of the object on each scale. The extraction process comprises performing convolution operation on input image I to obtain position feature maps in different scales, and generally using 4 scale feature maps F1,F2,F3,F4And the spectrum scale relation is that the lower characteristic spectrum is 2 times of the upper characteristic spectrum. And then fusing the upper layer map with the lower layer map. In the fusion process, firstly, the upper-layer features are sampled to the same scale of the lower-layer map, and then the feature map which is sampled and is subjected to element superposition with the lower-layer feature map to obtain four feature maps F which are subjected to fusion1’,F2’,F3’,F4. The characteristic pyramid structure is shown in FIG. 2 as a dotted box structure, where
Figure DEST_PATH_IMAGE001
Representing the superposition of the upper layer feature and the lower layer feature element. Arrows indicate the direction of feature map transfer. The location feature map is the basis for subsequent character positioning.
And sending the position feature map into the information attention module, and performing difference attention on the feature map by the information attention module to obtain accurate character position information.
Above information attention module
In the feature pyramid, the position information determined from top to bottom is accurate step by step, namely, the position coordinate acquired by the upper layer features has larger deviation relative to the actual character coordinate, and the character coordinate position deviation acquired by the lower layer features is smaller. By performing differential attention on the position information of different deviations, the character positioning information can be accurately obtained. The above information attention module is designed based on the theory.
Referring to fig. 3, the main processing flow of the module is as follows: firstly, a characteristic map F is matched1All characteristics of upper layer F2,F3,…FnPerforming upsampling to obtain an upsampled value1Feature profile Fu with the same dimensions2，Fu3…Fun. Then F is mixed1，Fu2,Fu3…FunThe element superposition is carried out to obtain a feature Fm, which is shown in the right parenthesis in fig. 3. Then carrying out convolution operation on Fm to obtain Fm ', Fm' scale and F1The size of the scale is the same, the number of channels is n, n is equal to the input F of the module1,F2,F3,…FnThe sum of the number of the characteristic spectrum channels is the same. And then, carrying out normalization processing on Fm' to obtain a difference attention degree score of the position characteristics, wherein the score represents the utilization rate of the position information contained in the position characteristic map. And then element multiplication is carried out on the normalized score and the input features to obtain a difference attention feature Fdot, and finally the feature Fdot is superposed to obtain character position information. The above information here represents all the characteristic information above this layer, in fig. 3
Figure 648857DEST_PATH_IMAGE001
The superposition of the characteristic elements is represented,
Figure 366277DEST_PATH_IMAGE002
representing the convolution operation and the arrows represent the direction of feature map transfer. The characteristic normalization formula is as follows:
Figure 151699DEST_PATH_IMAGE004
wherein p represents elements on the position characteristic map, q represents element values of the position characteristic map after normalization, x represents the abscissa position of the characteristic map, y represents the ordinate position of the characteristic map, c represents the number of channels of the characteristic map, and n represents the total number of channels of the characteristic map.
The above information attention module uses all feature information of the layer and the upper layer thereof, so that the influence of factors such as character distortion and illumination change on the features of a certain layer can be effectively weakened, the robustness of character positioning is improved, and the influence of conditions such as character distortion, blurring and brightness and chromaticity change on the positioning performance caused by the shooting angle problem is weakened.
Water level scale recognition module
This module mainly used discerns water level gauge scale. The characters on the water level ruler consist of a water level line and numbers, and the module firstly needs to identify the types of the numbers and the characters. The specific identification processing flow is as follows, firstly, the digital character image is deducted according to the character positioning information, and then the extracted image is sent to the depth network. Referring to fig. 4, the network consists of 4 convolutions and 2 fully-connected operations, and the input image is subjected to 4 convolution operations to obtain a feature F4Then F is added4Pulling into a characteristic vector, sending into full-connection operation to obtain a characteristic vector F5. Final eigenvector F5And obtaining the final character type output after the full connection operation is performed again. Here the characters output a total of 10 categories, representing the numbers 0 to 9 respectively. And determining the number of the current numeric character according to the class number output by the last network. In the drawings
Figure 169334DEST_PATH_IMAGE002
Representing convolution operation, fc representing full-join operation, arrows representing feature-map transfer direction, F1，F2,F3,F4Representation of a characteristic map, F5A feature vector is represented.
After the class identification of the digital characters is finished, the position information of the digital characters is sequenced, the digital characters at the lowest position are found, and finally, a water level scale value is confirmed according to the class information identified by the characters, wherein the value is the current water level height.
The embodiments of the present invention are only for illustrating the technical solutions of the present application, and the similar substitutes made by those skilled in the art on the basis of the present application shall all belong to the protection scope of the present invention:
1. the pyramid structure of the combined features herein may be replaced with other feature extraction mechanisms.
The above information is fused by using superposition, and other information fusion modes can be adopted instead.

Claims (2)

1. A water level ruler character positioning method based on the above information attention degree uses a position information extraction module, an above information attention degree module and a water level ruler scale identification module, and the specific method comprises the following steps:
the camera captures images of the water level gauge;
the position information extraction module extracts position characteristic information on different scales of the image to obtain a position characteristic map; superposing upper layer map information and lower layer map information in the position characteristic map and sending the superposed upper layer map information and lower layer map information to an upper information attention module for character positioning;
thirdly, the above information attention module performs differential attention on the position information captured by the local layer of features and all the upper layer of features in the position feature map to obtain accurate character position information; the specific treatment steps are as follows:
1) to the atlas F in the position characteristic atlas1All characteristic maps F of upper layer2,F3,…FnPerforming upsampling to obtain an upsampled value1With the same ruleCun characteristic spectrum Fu2，Fu3…Fun；
2) F is to be1，Fu2,Fu3…FunElement superposition is carried out to obtain characteristic Fm, convolution operation is carried out on Fm to obtain Fm ', the size of Fm' and F1The size of the scale is the same, the number of channels is n, n is equal to the input F of the module1,F2,F3,…FnThe sum of the number of the characteristic spectrum channels is the same;
3) normalizing Fm' to obtain a difference attention score of the position characteristics, wherein the score represents the utilization rate of position information contained in a position characteristic map;
4) the normalized score is compared to the feature profile F1，Fu2,Fu3…FunElement multiplication is carried out, and difference attention features Fdot are extracted;
5) superposing the characteristic Fdot to obtain character positioning information;
and (IV) using the character positioning information to carry out character scratching on the image, sending the scratched image into a character recognition module to carry out character recognition, and acquiring scale scales of the scale to complete output.
2. The water level ruler character positioning method based on the above information attention degree according to claim 1, wherein the above information attention degree module performs differential attention on the local layer of features and all the features on the upper layer in the position feature map to obtain accurate character position information.
CN202011413570.5A 2020-12-07 2020-12-07 Water level character positioning method based on information attention degree Active CN112329787B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011413570.5A CN112329787B (en) 2020-12-07 2020-12-07 Water level character positioning method based on information attention degree

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011413570.5A CN112329787B (en) 2020-12-07 2020-12-07 Water level character positioning method based on information attention degree

Publications (2)

Publication Number Publication Date
CN112329787A true CN112329787A (en) 2021-02-05
CN112329787B CN112329787B (en) 2023-06-30

Family

ID=74302523

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011413570.5A Active CN112329787B (en) 2020-12-07 2020-12-07 Water level character positioning method based on information attention degree

Country Status (1)

Country Link
CN (1) CN112329787B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040017944A1 (en) * 2002-05-24 2004-01-29 Xiaoging Ding Method for character recognition based on gabor filters
CN109766886A (en) * 2019-01-02 2019-05-17 南京林业大学 A kind of water level recognition methods based on image recognition
CN110705759A (en) * 2019-09-18 2020-01-17 平安科技（深圳）有限公司 Water level early warning monitoring method and device, storage medium and electronic equipment
CN112016556A (en) * 2020-08-21 2020-12-01 中国科学技术大学 Multi-type license plate recognition method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040017944A1 (en) * 2002-05-24 2004-01-29 Xiaoging Ding Method for character recognition based on gabor filters
CN109766886A (en) * 2019-01-02 2019-05-17 南京林业大学 A kind of water level recognition methods based on image recognition
CN110705759A (en) * 2019-09-18 2020-01-17 平安科技（深圳）有限公司 Water level early warning monitoring method and device, storage medium and electronic equipment
CN112016556A (en) * 2020-08-21 2020-12-01 中国科学技术大学 Multi-type license plate recognition method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
陈翠;刘正伟;陈晓生;骆曼娜;牛智星;阮聪;: "基于图像处理的水位信息自动提取技术", 水利信息化, no. 01 *

Also Published As

Publication number Publication date
CN112329787B (en) 2023-06-30

Similar Documents

Publication Publication Date Title
CN112818988B (en) Automatic identification reading method and system for pointer instrument
CN106355197A (en) Navigation image matching filtering method based on K-means clustering algorithm
CN112488046B (en) Lane line extraction method based on high-resolution images of unmanned aerial vehicle
CN109635726B (en) Landslide identification method based on combination of symmetric deep network and multi-scale pooling
CN111539330B (en) Transformer substation digital display instrument identification method based on double-SVM multi-classifier
Mostafa et al. Shadow identification in high resolution satellite images in the presence of water regions
CN112734729B (en) Water gauge water level line image detection method and device suitable for night light supplement condition and storage medium
CN110175556B (en) Remote sensing image cloud detection method based on Sobel operator
CN109635799B (en) Method for recognizing number of character wheel of gas meter
CN114266881A (en) Pointer type instrument automatic reading method based on improved semantic segmentation network
CN113837166B (en) Automatic pointer instrument reading method based on deep learning
CN115100615A (en) End-to-end lane line detection method based on deep learning
CN113657377B (en) Structured recognition method for mechanical bill image
CN116758045B (en) Surface defect detection method and system for semiconductor light-emitting diode
CN111476314B (en) Fuzzy video detection method integrating optical flow algorithm and deep learning
CN113269195A (en) Reading table image character recognition method and device and readable storage medium
CN112329787B (en) Water level character positioning method based on information attention degree
CN114926635B (en) Target segmentation method in multi-focus image combined with deep learning method
CN107330436B (en) Scale criterion-based panoramic image SIFT optimization method
CN116385477A (en) Tower image registration method based on image segmentation
CN111274858A (en) Business license identification method in network transaction supervision
CN114898347A (en) Machine vision identification method for pointer instrument
CN112381747A (en) Terahertz and visible light image registration method and device based on contour feature points
CN113554018B (en) Meter automatic modeling method based on machine vision
CN115359346B (en) Small micro-space identification method and device based on street view picture and electronic equipment

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