CN113487594A - Sub-pixel angular point detection method, system and medium based on deep learning - Google Patents
Sub-pixel angular point detection method, system and medium based on deep learning Download PDFInfo
- Publication number
- CN113487594A CN113487594A CN202110832567.5A CN202110832567A CN113487594A CN 113487594 A CN113487594 A CN 113487594A CN 202110832567 A CN202110832567 A CN 202110832567A CN 113487594 A CN113487594 A CN 113487594A
- Authority
- CN
- China
- Prior art keywords
- sub
- corner
- pixel
- angular point
- coordinates
- 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
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 66
- 238000013135 deep learning Methods 0.000 title claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 29
- 238000000034 method Methods 0.000 claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 19
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 7
- 239000003550 marker Substances 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 abstract description 3
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 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
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a sub-pixel angular point detection method, a system and a medium based on deep learning, which relate to the technical field of digital image processing, and the method comprises the following steps: step 1: acquiring training data of a subpixel level to train a subpixel corner detection network; step 2: acquiring an edge subgraph; and step 3: and inputting the edge sub-image into a sub-pixel corner detection network. The method can overcome the problems of low detection precision and poor generalization performance of pixel corner detection, thereby achieving the beneficial effect of quickly and accurately realizing sub-pixel corner segmentation.
Description
Technical Field
The invention relates to the technical field of digital image processing, in particular to a sub-pixel corner detection method, a system and a medium based on deep learning.
Background
In many fields at present, a method for extracting sub-pixel level feature point coordinates from a digital image is urgently needed, for example, in the field of medical image analysis, sub-pixel level cell edge coordinates are often needed to be extracted; in the surveying and mapping field, if accurate sub-pixel level coordinates of some characteristic points can be extracted through a digital image of a building, the information such as the height of the building can be directly measured through the digital image, and therefore manpower and material resources can be greatly saved.
There are related documents f.zhao, c.wei, j.wang, and j.tang, "An Automated X-corner Detection Algorithm (AXDA)," JSW, vol.6, No.5, pp.791-797,2011, which describe a method of tracking the light intensity change from dark to light in An image, the X angular center being calculated using the minimum correlation coefficient method.
For the prior art, in the related technical field, the existing feature point detection algorithm at the pixel level and the sub-pixel level has low detection precision, small application range, limitation to a certain specific problem and poor generalization performance.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a sub-pixel corner detection method, a system and a medium based on deep learning, which can solve the problems of low detection precision and poor generalization performance of pixel corner detection so as to achieve the effect of quickly and accurately realizing sub-pixel corner segmentation.
According to the sub-pixel corner detection method, the system and the medium provided by the invention, the scheme is as follows:
in a first aspect, a deep learning-based sub-pixel corner detection method is provided, and the method includes:
acquiring training data of a subpixel level to train a subpixel corner detection network;
acquiring an edge subgraph;
and inputting the edge sub-image into a sub-pixel corner detection network.
Preferably, the manner of acquiring training data at the sub-pixel level includes:
acquiring corner coordinates on an original picture by using an openCV sub-pixel corner detection algorithm;
reducing the original picture by a certain proportion, wherein the obtained angular point coordinates can also change in an equal proportion, and the changed angular point coordinates are approximately used as actual coordinates of an angular point on the low-precision picture and are used as real marks for training;
and (4) taking the low-precision picture as input, and taking the corner point coordinates acquired on the high-precision picture as a mark to train a neural network.
Preferably, the manner of acquiring the edge subgraph includes:
acquiring an angular point region by applying a Harris algorithm on the image;
taking the centroid of each region as a preselected corner point coordinate;
after acquiring the coordinates of preselected angular points, taking sub-images with the size of 15 multiplied by 15 around each preselected angular point;
and processing the obtained subgraph by using a Canny operator to obtain an edge subgraph.
Preferably, the sub-pixel corner detection network comprises:
solving the regression problem of continuous values by using a sub-pixel angular point detection network, inputting the obtained edge sub-images, and outputting the predicted offset of the sub-pixel-level angular point coordinates relative to the pre-selected angular point coordinates;
the training uses the square of the distance of the predicted corner from the corner in the marker data as a loss function.
In a second aspect, a deep learning-based sub-pixel corner detection system is provided, the system comprising:
the data module is used for acquiring training data of a sub-pixel level and training a sub-pixel corner detection network;
the image module is used for acquiring an edge subgraph;
and the detection module inputs the edge sub-image into a sub-pixel corner detection network.
Preferably, the data module includes:
acquiring corner coordinates on an original picture by using an openCV sub-pixel corner detection algorithm;
reducing the original picture by a certain proportion, wherein the obtained angular point coordinates can also change in an equal proportion, and the changed angular point coordinates are approximately used as actual coordinates of an angular point on the low-precision picture and are used as real marks for training;
and (4) taking the low-precision picture as input, and taking the corner point coordinates acquired on the high-precision picture as a mark to train a neural network.
Preferably, the image module includes:
acquiring an angular point region by applying a Harris algorithm on the image;
taking the centroid of each region as a preselected corner point coordinate;
after acquiring the coordinates of preselected angular points, taking sub-images with the size of 15 multiplied by 15 around each preselected angular point;
and processing the obtained subgraph by using a Canny operator to obtain an edge subgraph.
Preferably, the detection module includes:
solving the regression problem of continuous values by using a sub-pixel angular point detection network, inputting the obtained edge sub-images, and outputting the predicted offset of the sub-pixel-level angular point coordinates relative to the pre-selected angular point coordinates;
the training uses the square of the distance of the predicted corner from the corner in the marker data as a loss function.
In a third aspect, a computer-readable storage medium is provided, in which a computer program is stored, which, when being executed by a processor, performs the steps in the deep learning-based sub-pixel corner detection method.
Compared with the prior art, the invention has the following beneficial effects:
1. by adopting the neural network structure, the detection precision and the generalization performance of sub-pixel angular point detection are improved; 2. by adopting the neural network structure, the effect of quickly and accurately realizing the sub-pixel angular points is achieved.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:
FIG. 1 is a schematic flow chart of a sub-pixel corner detection method;
fig. 2 is a schematic diagram of a network structure of a sub-pixel corner detection method.
Detailed Description
The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that it would be obvious to those skilled in the art that various changes and modifications can be made without departing from the spirit of the invention. All falling within the scope of the present invention.
The embodiment of the invention provides a sub-pixel corner detection method based on deep learning, and as shown in figure 1, the method comprises the steps of firstly obtaining sub-pixel-level training data to train a sub-pixel corner detection network, then obtaining an edge sub-image by applying a specific algorithm and a processing mode, and finally using the sub-pixel corner detection network. When the sub-pixel level training data is obtained, the pixel level corner points on the high-precision picture also correspond to the sub-pixel level corner points on the low-precision picture, and along with the rise of picture precision, the corner point coordinates obtained by a classical algorithm can continuously approach the real corner point coordinates.
Referring to fig. 2, we can obtain training data at the sub-pixel level as follows: firstly, acquiring corner coordinates on an original picture to be processed by using a subpixel corner detection algorithm of openCV, reducing the processed original picture by a certain proportion after the corner coordinates are obtained by algorithm calculation, wherein the proportion can be selected according to actual requirements, the previously acquired corner coordinates can also change in an equal proportion after the original picture is reduced, then the corner coordinates are approximately taken as actual coordinates of corners on a low-precision picture, and the corner coordinates which change in an equal proportion are taken as real marks for training; because the low-precision pictures are only visible to the program, the low-precision pictures can be used as input, and at the moment, the corner point coordinates acquired on the high-precision pictures can be used as the mark training neural network.
Secondly, we need to obtain edge sub-images, sub-pixel level coordinates of the corner point are only related to sub-images with certain sizes around the corner point, and the relationship between pixel points too far away from the corner point and the coordinates of the corner point is very small. Firstly, acquiring corner regions by applying a Harris algorithm on a reduced original picture, acquiring the corner regions, then taking the centroid of each region as a preselected corner coordinate, acquiring the preselected corner coordinate, and then selecting sub-pictures with the size of 15 multiplied by 15 around each preselected corner; and processing the obtained subgraph by using a Canny operator to finally obtain the required edge subgraph.
Finally, a sub-pixel corner detection network is used, the regression problem of continuous values is solved through the sub-pixel corner detection network, the obtained edge sub-images are input, an output result can be obtained, and the output is the offset of the predicted sub-pixel level corner coordinates relative to the pre-selection corner coordinates. Because the input dimension required to be processed by the neural network is very small, and the common input dimension is generally selected to be 15 × 15 × 3, the neural network architecture is not too complex when being considered, otherwise, the phenomenon of overfitting is easy to occur, and the network architecture is 1-layer convolution + 1-layer full connection, so that the detection precision of the network architecture reaches the highest. And during training, the square of the distance between the predicted corner and the corner in the marking data is used as a loss function, the reason for the mode is to hopefully minimize the average distance between the predicted corner and the real mark and keep the training function consistent with the target.
The embodiment of the invention provides a sub-pixel corner detection method based on deep learning, which aims at sub-pixel corner detection, firstly applies a method based on deep learning, proposes the idea of taking an edge image as network input, obtains an optimized network structure through multiple experiments, surpasses the precision of the conventional sub-pixel corner detection algorithm, and has better generalization.
The embodiment of the invention also provides a sub-pixel corner detection system based on deep learning, which comprises a data module, an image module and a detection module. The data module is mainly used for acquiring training data of a sub-pixel level to train a sub-pixel corner detection network, a pixel-level corner on a high-precision picture also corresponds to a sub-pixel-level corner on a low-precision picture, and along with the rise of picture precision, a corner coordinate acquired by a classical algorithm can continuously approach a real corner coordinate.
When training data is obtained, obtaining corner coordinates on an original picture to be processed by using a sub-pixel corner detection algorithm of openCV, reducing the original picture to a certain proportion, wherein the proportion can be set according to actual requirements, after the picture is reduced, the corner coordinates obtained in the previous step can be changed in an equal proportion, the changed corner coordinates are approximately used as actual coordinates of the corner coordinates on the low-precision picture, and the changed corner coordinates are used as real marks for training. Because the low-precision pictures are only visible to the program, the low-precision pictures can be used as input, and at the moment, the corner point coordinates acquired on the high-precision pictures can be used as the mark training neural network.
The image module is used for acquiring an edge sub-picture, and in the process of acquiring the edge sub-picture, sub-pixel level coordinates of a corner point are only related to the sub-picture with a certain size around the corner point, and the relation between pixel points far away from the corner point and the coordinates of the corner point is very small. Firstly, a Harris algorithm is applied to an original image to be processed to obtain corner regions, and after the corner regions are obtained, the centroid in each region is used as a preselected corner coordinate. After the preselected corner point coordinates are obtained, sub-images with the size of 15 multiplied by 15 around each preselected corner point are selected, and finally the obtained sub-images are processed by a Canny operator, so that the required edge sub-images can be obtained.
In the detection module, the regression problem of the continuous values can be solved by using a sub-pixel corner detection network, the edge sub-images obtained in the previous step are input, and the offset of the predicted sub-pixel level corner coordinates relative to the pre-selected corner coordinates is output. Because the input dimension required to be processed by the neural network is very small, and the general input dimension is selected to be 15 × 15 × 3 under normal conditions, the neural network architecture is not too complex when being considered, otherwise, the phenomenon of overfitting is easy to occur, and the network architecture is 1-layer convolution and + 1-layer full connection, so that the detection precision is highest. The square of the distance between the predicted corner and the corner in the marker data is used as a loss function in the training process, because it is desirable to minimize the average distance between the predicted corner and the true marker, and the training function is consistent with the target.
The embodiment of the invention provides a sub-pixel angular point detection system based on deep learning, which solves the problems of low detection precision and poor generalization performance of sub-pixel angular point detection by mutually matching a data module, an image module and a detection module in sequence and adopting a neural network structure, and achieves the effect of quickly and accurately realizing sub-pixel angular point segmentation.
Those skilled in the art will appreciate that, in addition to implementing the system and its various devices, modules, units provided by the present invention as pure computer readable program code, the system and its various devices, modules, units provided by the present invention can be fully implemented by logically programming method steps in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Therefore, the system and various devices, modules and units thereof provided by the invention can be regarded as a hardware component, and the devices, modules and units included in the system for realizing various functions can also be regarded as structures in the hardware component; means, modules, units for performing the various functions may also be regarded as structures within both software modules and hardware components for performing the method.
The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention. The embodiments and features of the embodiments of the present application may be combined with each other arbitrarily without conflict.
Claims (9)
1. A sub-pixel corner detection method based on deep learning is characterized by comprising the following steps:
step 1: acquiring training data of a subpixel level to train a subpixel corner detection network;
step 2: acquiring an edge subgraph;
and step 3: and inputting the edge sub-image into a sub-pixel corner detection network.
2. The method of claim 1, wherein step 1 comprises:
step 1-1: acquiring corner coordinates on an original picture by using an openCV sub-pixel corner detection algorithm;
step 1-2: reducing the original picture by a certain proportion, wherein the obtained angular point coordinates can also change in an equal proportion, and the changed angular point coordinates are approximately used as actual coordinates of an angular point on the low-precision picture and are used as real marks for training;
step 1-3: and (4) taking the low-precision picture as input, and taking the corner point coordinates acquired on the high-precision picture as a mark to train a neural network.
3. The method of claim 1, wherein step 2 comprises:
step 2-1: acquiring an angular point region by applying a Harris algorithm on the image;
step 2-2: taking the centroid of each region as a preselected corner point coordinate;
step 2-3: after acquiring the coordinates of preselected angular points, taking sub-images with the size of 15 multiplied by 15 around each preselected angular point;
step 2-4: and processing the obtained subgraph by using a Canny operator to obtain an edge subgraph.
4. The method of claim 1, wherein step 3 comprises:
step 3-1: solving the regression problem of continuous values by using a sub-pixel angular point detection network, inputting the obtained edge sub-images, and outputting the predicted offset of the sub-pixel-level angular point coordinates relative to the pre-selected angular point coordinates;
step 3-2: the training uses the square of the distance of the predicted corner from the corner in the marker data as a loss function.
5. A deep learning-based sub-pixel corner detection system, the system comprising:
the data module is used for acquiring training data of a sub-pixel level and training a sub-pixel corner detection network;
the image module is used for acquiring an edge subgraph;
and the detection module inputs the edge sub-image into a sub-pixel corner detection network.
6. The system of claim 5, wherein the data module comprises:
acquiring corner coordinates on an original picture by using an openCV sub-pixel corner detection algorithm;
reducing the original picture by a certain proportion, wherein the obtained angular point coordinates can also change in an equal proportion, and the changed angular point coordinates are approximately used as actual coordinates of an angular point on the low-precision picture and are used as real marks for training;
and (4) taking the low-precision picture as input, and taking the corner point coordinates acquired on the high-precision picture as a mark to train a neural network.
7. The system of claim 5, wherein the image module comprises:
acquiring an angular point region by applying a Harris algorithm on the image;
taking the centroid of each region as a preselected corner point coordinate;
after acquiring the coordinates of preselected angular points, taking sub-images with the size of 15 multiplied by 15 around each preselected angular point;
and processing the obtained subgraph by using a Canny operator to obtain an edge subgraph.
8. The system of claim 5, wherein the detection module comprises:
solving the regression problem of continuous values by using a sub-pixel angular point detection network, inputting the obtained edge sub-images, and outputting the predicted offset of the sub-pixel-level angular point coordinates relative to the pre-selected angular point coordinates;
the training uses the square of the distance of the predicted corner from the corner in the marker data as a loss function.
9. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110832567.5A CN113487594B (en) | 2021-07-22 | 2021-07-22 | Sub-pixel corner detection method, system and medium based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110832567.5A CN113487594B (en) | 2021-07-22 | 2021-07-22 | Sub-pixel corner detection method, system and medium based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113487594A true CN113487594A (en) | 2021-10-08 |
CN113487594B CN113487594B (en) | 2023-12-01 |
Family
ID=77943355
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110832567.5A Active CN113487594B (en) | 2021-07-22 | 2021-07-22 | Sub-pixel corner detection method, system and medium based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113487594B (en) |
Citations (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU2094843C1 (en) * | 1992-12-04 | 1997-10-27 | Георгий Алексеевич Легков | Device which models neural network |
CN101465002A (en) * | 2009-01-05 | 2009-06-24 | 东南大学 | Method for orientating secondary pixel edge of oval-shaped target |
US20100254611A1 (en) * | 2009-03-30 | 2010-10-07 | Carl Zeiss Sms Gmbh | Method and device for determining the position of an edge of a marker structure with subpixel accuracy in an image, having a plurality of pixels, of the marker structure |
CN103345755A (en) * | 2013-07-11 | 2013-10-09 | 北京理工大学 | Chessboard angular point sub-pixel extraction method based on Harris operator |
CN103927750A (en) * | 2014-04-18 | 2014-07-16 | 上海理工大学 | Detection method of checkboard grid image angular point sub pixel |
US20160140713A1 (en) * | 2013-07-02 | 2016-05-19 | Guy Martin | System and method for imaging device modelling and calibration |
CN105654473A (en) * | 2015-12-25 | 2016-06-08 | 傅之成 | Sub-pixel edge detection method |
CN106529548A (en) * | 2016-10-24 | 2017-03-22 | 南京工业大学 | Sub-pixel level multi-scale Harris corner detection algorithm |
CN108257185A (en) * | 2018-01-03 | 2018-07-06 | 上海兴芯微电子科技有限公司 | More checkerboard angle point detection process and camera marking method |
CN108399412A (en) * | 2018-03-07 | 2018-08-14 | 鞍钢集团矿业有限公司 | A kind of X-type angular-point sub-pixel extracting method |
CN108895959A (en) * | 2018-04-27 | 2018-11-27 | 电子科技大学 | A kind of camera calibration plate angle point calculating method based on sub-pix |
CN109118473A (en) * | 2018-07-03 | 2019-01-01 | 深圳大学 | Angular-point detection method, storage medium and image processing system neural network based |
CN110020989A (en) * | 2019-05-23 | 2019-07-16 | 西华大学 | A kind of depth image super resolution ratio reconstruction method based on deep learning |
CN110033469A (en) * | 2019-04-01 | 2019-07-19 | 北京科技大学 | A kind of sub-pixel edge detection method and system |
CN110675320A (en) * | 2019-09-24 | 2020-01-10 | 南京工程学院 | Method for sharpening target image under spatial parameter change and complex scene |
CN111243032A (en) * | 2020-01-10 | 2020-06-05 | 大连理工大学 | Full-automatic checkerboard angular point detection method |
US20200264082A1 (en) * | 2017-10-18 | 2020-08-20 | Suzhou H-C Soil & Water Science And Technology Co., Ltd | Image measurement device and method for the surface deformation of specimen based on sub-pixel corner detection |
-
2021
- 2021-07-22 CN CN202110832567.5A patent/CN113487594B/en active Active
Patent Citations (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU2094843C1 (en) * | 1992-12-04 | 1997-10-27 | Георгий Алексеевич Легков | Device which models neural network |
CN101465002A (en) * | 2009-01-05 | 2009-06-24 | 东南大学 | Method for orientating secondary pixel edge of oval-shaped target |
US20100254611A1 (en) * | 2009-03-30 | 2010-10-07 | Carl Zeiss Sms Gmbh | Method and device for determining the position of an edge of a marker structure with subpixel accuracy in an image, having a plurality of pixels, of the marker structure |
US20160140713A1 (en) * | 2013-07-02 | 2016-05-19 | Guy Martin | System and method for imaging device modelling and calibration |
CN103345755A (en) * | 2013-07-11 | 2013-10-09 | 北京理工大学 | Chessboard angular point sub-pixel extraction method based on Harris operator |
CN103927750A (en) * | 2014-04-18 | 2014-07-16 | 上海理工大学 | Detection method of checkboard grid image angular point sub pixel |
CN105654473A (en) * | 2015-12-25 | 2016-06-08 | 傅之成 | Sub-pixel edge detection method |
CN106529548A (en) * | 2016-10-24 | 2017-03-22 | 南京工业大学 | Sub-pixel level multi-scale Harris corner detection algorithm |
US20200264082A1 (en) * | 2017-10-18 | 2020-08-20 | Suzhou H-C Soil & Water Science And Technology Co., Ltd | Image measurement device and method for the surface deformation of specimen based on sub-pixel corner detection |
CN108257185A (en) * | 2018-01-03 | 2018-07-06 | 上海兴芯微电子科技有限公司 | More checkerboard angle point detection process and camera marking method |
CN108399412A (en) * | 2018-03-07 | 2018-08-14 | 鞍钢集团矿业有限公司 | A kind of X-type angular-point sub-pixel extracting method |
CN108895959A (en) * | 2018-04-27 | 2018-11-27 | 电子科技大学 | A kind of camera calibration plate angle point calculating method based on sub-pix |
CN109118473A (en) * | 2018-07-03 | 2019-01-01 | 深圳大学 | Angular-point detection method, storage medium and image processing system neural network based |
CN110033469A (en) * | 2019-04-01 | 2019-07-19 | 北京科技大学 | A kind of sub-pixel edge detection method and system |
CN110020989A (en) * | 2019-05-23 | 2019-07-16 | 西华大学 | A kind of depth image super resolution ratio reconstruction method based on deep learning |
CN110675320A (en) * | 2019-09-24 | 2020-01-10 | 南京工程学院 | Method for sharpening target image under spatial parameter change and complex scene |
CN111243032A (en) * | 2020-01-10 | 2020-06-05 | 大连理工大学 | Full-automatic checkerboard angular point detection method |
Non-Patent Citations (5)
Title |
---|
THANH HONG-PHUOC, ET AL.,: ""A scale and Rotational Invariant Key-point Detector based on Sparse Coding"", 《WWW.ARXIV.ORG/PDF/2010.06264.PDF》 * |
刘小勇;宫岩;李荣丽;郝兆朋;崔高健;王晖;范依航: "基于BP神经网络的数字图像相关非迭代灰度梯度算法", 机床与液压, no. 001 * |
卢英;王慧琴;佟威;李俊杰;: "基于Harris-张正友平面标定法的摄像机标定算法", 西安建筑科技大学学报(自然科学版), no. 06 * |
李明金;熊显名;张绍兵;: "一种基于OpenCV的摄像机标定新方法", 激光与光电子学进展, no. 12 * |
陶雪焦;汪亚明;郑俊褒;: "改进的多尺度Harris亚像素角点检测", 工业控制计算机, no. 03 * |
Also Published As
Publication number | Publication date |
---|---|
CN113487594B (en) | 2023-12-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10803554B2 (en) | Image processing method and device | |
US20180137321A1 (en) | Method and system for decoding two-dimensional code using weighted average gray-scale algorithm | |
CN108986152B (en) | Foreign matter detection method and device based on difference image | |
CN111915483B (en) | Image stitching method, device, computer equipment and storage medium | |
US20110228124A1 (en) | Character recognition preprocessing method and apparatus | |
CN108074237B (en) | Image definition detection method and device, storage medium and electronic equipment | |
CN112930677B (en) | Method for switching between first lens and second lens and electronic device | |
CN111311482A (en) | Background blurring method and device, terminal equipment and storage medium | |
Wang et al. | Noise-robust color edge detector using gradient matrix and anisotropic Gaussian directional derivative matrix | |
CN108764139B (en) | Face detection method, mobile terminal and computer readable storage medium | |
CN112991218B (en) | Image processing method, device, equipment and storage medium | |
CN104809716A (en) | Method and equipment for determining image offset | |
KR101440826B1 (en) | Image registration method and apparatus | |
CN110689014B (en) | Method and device for detecting region of interest, electronic equipment and readable storage medium | |
CN109166075A (en) | One kind being directed to small overlapping region image split-joint method | |
CN111882588B (en) | Image block registration method and related product | |
CN111598917A (en) | Data embedding method, device, equipment and computer readable storage medium | |
CN111563517A (en) | Image processing method, image processing device, electronic equipment and storage medium | |
CN113487594A (en) | Sub-pixel angular point detection method, system and medium based on deep learning | |
CN108470326B (en) | Image completion method and device | |
JP2009146150A (en) | Method and device for detecting feature position | |
CN115619678A (en) | Image deformation correction method and device, computer equipment and storage medium | |
CN113837998B (en) | Method and device for automatically adjusting and aligning pictures based on deep learning | |
CN111179166B (en) | Image processing method, device, equipment and computer readable storage medium | |
CN107369150B (en) | Method for detecting rectangular target and rectangular target detection device |
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 |