CN111274897A - Method and device for identifying mark point image - Google Patents
Method and device for identifying mark point image Download PDFInfo
- Publication number
- CN111274897A CN111274897A CN202010042272.3A CN202010042272A CN111274897A CN 111274897 A CN111274897 A CN 111274897A CN 202010042272 A CN202010042272 A CN 202010042272A CN 111274897 A CN111274897 A CN 111274897A
- Authority
- CN
- China
- Prior art keywords
- image
- images
- equivalent
- sequence
- short
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/48—Matching video sequences
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/49—Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
Abstract
The application relates to a method and a device for identifying a mark point image. The method comprises the following steps: acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points, and selecting partial images from the image sequence of the mark points as reference images; selecting at least one group of short image sequences from the image sequences of the mark points; equating the short image sequence to an equivalent image; according to the gray-scale invariant model, the displacement of the equivalent image relative to the reference image is obtained, and the problem of low identification precision of the mark point image is solved, so that the influence of time and space inconsistency of image gray scale on the identification precision of the mark point image is reduced, and the identification precision of the mark point image can meet the technical requirements of related application fields.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a method and a device for recognizing a mark point image.
Background
With the development of image recognition technology, digital image recognition technology based on mark points has been widely applied to the fields of medical feature positioning, Artificial Intelligence (AI) control, engineering structure deformation measurement and the like, thus promoting the development of various application fields and bringing great convenience to life.
In the related art, the identification precision of the mark point image is easily affected by the time and space inconsistency of the image gray scale, so that the existing mark point image identification technology cannot meet the technical requirements of the related application field on the identification precision.
Aiming at the problem of low identification precision of the mark point image in the related technology, no effective solution is provided at present.
Disclosure of Invention
The invention provides a method and a device for identifying a mark point image, aiming at the problem of low identification precision of the mark point image in the related art, and at least solving the problem.
According to an aspect of the present invention, there is provided a method of landmark image recognition, comprising the steps of:
acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points, and selecting partial images from the image sequence of the mark points as reference images;
selecting at least one group of short image sequences from the image sequences of the mark points;
equating the short image sequence to an equivalent image;
and obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model.
In one embodiment, the selecting at least one group of short image sequences from the image sequences of the marker points includes:
and selecting a plurality of groups of short image sequences of the mark points from the image sequences of the mark points, wherein the difference between two adjacent groups of short image sequences is a preset image quantity length.
In one embodiment, the equivalent of the sequence of short images into equivalent images comprises:
and respectively equating the plurality of groups of short image sequences into a plurality of equivalent images, and taking the equivalent images as equivalent image sequences.
In one embodiment, the obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model includes:
and matching the reference image with the equivalent image sequence according to a gray-scale invariant model to obtain the equivalent image sequence and the deformation field data of the displacement of the reference image.
In one embodiment, the selecting at least one group of short image sequences from the image sequences of the marker points includes:
and selecting a group of short image sequences in the image sequences of the mark points according to a preset sequence length.
In one embodiment, the obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model includes:
obtaining the displacement of the equivalent image of the selected short image sequence relative to the reference image according to the gray-scale invariant model;
taking the area where the short image sequence is located as a sequence selection area, translating the sequence selection area by a preset image quantity length in the time direction of the image sequence of the mark point, updating the short image sequence, and calculating corresponding displacement until all the image sequences of the mark point are selected;
and acquiring deformation field data according to all the calculated displacements.
In one embodiment, the model based on gray scale invariance comprises:
wherein X is the displacement value of the mark point, U is the image displacement of the mark point, α is the related parameter, delta t is the time interval between two images, FiRepresenting the gray value of the reference image, GiThe gray value of the deformed image is shown, and m represents m frames of images before and after the ith frame.
In one embodiment, the acquiring a plurality of successively acquired images of the landmark points includes:
and selecting an image subregion containing the mark point to obtain a plurality of continuously acquired images of the image subregion.
In one embodiment, the selecting a partial image from the image sequence of the marker point as a reference image includes:
continuously selecting a plurality of images in the image sequence of the mark point;
and equivalently converting the selected multiple images into a reference image.
According to another aspect of the present invention, there is also provided an apparatus for landmark image recognition, the apparatus including:
the image acquisition module is used for acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points;
the image selection module is used for selecting partial images from the image sequences of the mark points as reference images and selecting at least one group of short image sequences;
the image equivalence module is used for enabling the short image sequence to be equivalent to an equivalent image;
and the displacement solving module is used for obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model.
The method and the device for identifying the marker point image acquire a plurality of continuously acquired images of the marker point to form an image sequence of the marker point, and select a partial image from the image sequence of the marker point as a reference image; selecting at least one group of short image sequences from the image sequences of the mark points; the short image sequence is equivalent to an equivalent image; according to the gray-scale invariant model, the displacement of the equivalent image relative to the reference image is obtained, and the problem of low identification precision of the mark point image is solved, so that the influence of time and space inconsistency of image gray scale on the identification precision of the mark point image is reduced, and the identification precision of the mark point image can meet the technical requirements of related application fields.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
FIG. 1 is a flow chart of a method of landmark image identification in an embodiment of the present invention;
FIG. 2 is a schematic diagram of a sequence of images simulating the vibration process of a landmark point in an embodiment in accordance with the invention;
FIG. 3 is a first schematic diagram of a displacement-time curve of a marker point measured according to various methods;
FIG. 4 is a schematic illustration of random and systematic errors of displacement measurements according to different methods;
FIG. 5 is a schematic illustration of uniaxial tensile experiments with index points in an example according to the invention;
FIG. 6 is a second schematic view of a displacement-time curve of a marker point measured according to a different method;
FIG. 7 is a third schematic representation of a strain-time curve of a marker point measured according to a different method;
FIG. 8 is a schematic diagram of an experimental setup for truss motion measurement of a deployable antenna in an embodiment of the invention;
FIG. 9 is a schematic illustration of the distance between two marker points measured according to a different method;
fig. 10 is a block diagram of the configuration of an apparatus for marker point image recognition according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
It should be noted that the terms "first", "second" and "third" related to the embodiments of the present invention only distinguish similar objects, and do not represent specific ordering for the objects, and the terms "first", "second" and "third" may be interchanged with specific order or sequence, where permitted. It is to be understood that the terms "first," "second," and "third" are used interchangeably where appropriate to enable embodiments of the present invention described herein to be practiced in sequences other than those illustrated or described herein.
The method for identifying the mark point image can be applied to the fields of medical characteristic positioning, artificial intelligence AI control, engineering structure deformation measurement and the like, such as real-time monitoring of motion deformation of large bridges, geotechnical structures and aerospace large deployable structures.
In one embodiment, fig. 1 is a flowchart of a method for landmark image recognition according to an embodiment of the present invention, and as shown in fig. 1, there is provided a method for landmark image recognition, including the steps of:
step S110, acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points, and selecting partial images from the image sequence of the mark points as reference images;
wherein, from ti-mTime ti+mAt that moment, the image acquisition device acquires 2m +1 consecutive images of the marker point, constituting an image sequence of the marker point.
It should be noted that the mark points may be positions to be monitored in medical feature positioning, artificial intelligence AI control and engineering structure deformation measurement, a monitoring area is selected as the mark points according to the requirements of practical application, a plurality of mark point images within a time period to be monitored are acquired through image acquisition equipment, and the acquisition number of the mark point images can be determined according to the requirements of practical research and cost calculation.
Step S120, at least one group of short image sequences are selected from the image sequences of the mark points.
The short image sequence is a marker image sequence composed of a plurality of marker images.
Step S130, the short image sequence is equivalent to an equivalent image.
It should be noted that a marker image sequence formed by a plurality of marker images is equivalent to a high-quality equivalent image according to the following formula:
wherein X is the displacement value of the mark point, and U is the mark pointImage shift, f (X) representing the gray value of the equivalent image, gtThe gray value of the deformed image is shown, and m represents m frames of images before and after the ith frame.
And step S140, acquiring the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model.
It should be noted that, according to the gray-scale invariant model, the reference image and the equivalent image are subjected to matching analysis to obtain a displacement function of the equivalent image; simplifying the gray scale invariant model, carrying out Taylor expansion on the displacement function, and carrying out simplification operation on the displacement function according to the digital image correlation DIC of the mark point to obtain the displacement function related to time, displacement and related parameters; and solving the related parameters of the displacement function according to the simplified solving displacement of the gray scale invariant model and a Newton-Raphson iteration method to obtain the displacement of the equivalent image relative to the reference image.
The displacement of the displacement function refers to the lateral displacement value and the longitudinal displacement value of the mark point, and the solid deformation is considered to have continuity in time, so that the displacement function of the mark point is a continuous function with respect to time, and for digital image correlation DIC of the mark point, the deformation of the mark point is not considered, and only the rigid displacement of the mark point is considered.
It is further noted that the correlation of digital images with the introduction of temporal continuity can establish the equations for 2m +1 gray scale invariant assumptions:
wherein f (x) represents the gray value of the reference image, gtGray values representing the deformed image;
and according to the gray-scale invariant model, performing matching analysis on the reference image and the equivalent image to obtain displacement data of each point on the surface of the equivalent image, and generating deformation field data of the displacement of the equivalent image sequence and the reference image so as to obtain a displacement function of the equivalent image sequence.
Considering that the solid deformation has continuity in time, therefore, the displacement function U (X, t) of the marker point is a continuous function with respect to time, and the function U (X, t) is subjected to a taylor expansion, which can be expressed as:
where Δ t is the time interval between two images.
For the digital image correlation of the mark point, the deformation of the mark point itself is not considered, only the rigid displacement of the mark point is considered, therefore, the displacement can be simplified as follows:
wherein the content of the first and second substances,
finally, expressing the marker point digital correlation matching algorithm considering time continuity by using the gray-scale invariant assumption can be expressed as:
α in the equation can be solved by a Newton-Raphson method, and mu and v, namely displacement values of the marking points in the transverse direction and the longitudinal direction are obtained through optimization solution.
According to the method for identifying the mark point image, at least one group of short image sequences are selected from the image sequences of the mark points, the short image sequences are equivalent to equivalent images, the displacement of the equivalent images relative to the reference image is obtained according to the gray scale invariant model, the time domain noise of the mark point image is reduced, and the problem of low identification precision of the mark point image is solved, so that the influence of time and space inconsistency of the gray scale of the mark point image on the identification precision of the mark point image is reduced, the identification precision of the mark point image is finally improved, and the identification precision of the mark point image can meet the technical requirements of related application fields.
In one embodiment, a method for landmark image recognition is provided, and step S120 includes step S220:
step S220, selecting multiple groups of short image sequences of the mark points from the image sequences of the mark points, wherein the difference between two adjacent groups of short image sequences is a preset image quantity length.
The example is illustrated in which 9 images are acquired in succession: from t0Time t8At the moment, the image acquisition equipment acquires 9 continuous images of the mark point to form an image sequence of the mark point, the sequence length of a preset short image sequence is 3, a first group of short image sequences are selected from the image sequence, the images of the first group of short image sequences are respectively P1, P2 and P3, a second group of short image sequences are selected from the image sequence, the images of the second group of short image sequences are respectively P2, P3 and P4, and the number length of the preset images is 1; a first group of short image sequences is selected from the image sequences, the images of the first group of short image sequences are respectively P1, P2 and P3, a second group of short image sequences is selected from the image sequences, the images of the second group of short image sequences are respectively P3, P4 and P5, and the preset number of images is 2 in length.
The short image sequence refers to a marker image sequence composed of a plurality of marker images; the marking points can be positions to be monitored in medical characteristic positioning, artificial intelligence AI control and engineering structure deformation measurement, monitoring areas are selected as the marking points according to the requirements of practical application, a plurality of marking point images in a time period needing to be monitored are collected through image collecting equipment, and the collecting quantity of the marking point images can be determined according to the requirements of practical research and cost calculation.
According to the method for identifying the mark point image, multiple groups of short image sequences of the mark points are selected from the image sequences of the mark points, displacement of an equivalent image relative to a reference image is obtained according to a gray-scale invariant model, time domain noise of the mark point image is reduced, and the problem of low identification precision of the mark point image is solved, so that the influence of time and space inconsistency of the gray scale of the mark point image on the identification precision of the mark point image is reduced, the identification precision of the mark point image is improved, and the identification precision of the mark point image can meet the technical requirements of related application fields.
In one embodiment, a method for identifying a landmark image is provided, and the step S130 includes the step S330:
and step S330, respectively equating the multiple groups of short image sequences into multiple equivalent images, and taking the multiple equivalent images as equivalent image sequences.
It should be noted that a marker image sequence composed of a plurality of marker images is equivalent to one equivalent image with high quality.
By the method for identifying the mark point image, a plurality of groups of short image sequences are respectively equivalent into a plurality of equivalent images, the equivalent images are used as equivalent image sequences, displacement of the equivalent images relative to a reference image is obtained according to a gray-scale invariant model, time domain noise of the mark point image is reduced, and the problem of low identification precision of the mark point image is solved, so that the influence of time and space inconsistency of the gray scale of the mark point image on the identification precision of the mark point image is reduced, the identification precision of the mark point image is finally improved, and the identification precision of the mark point image can meet the technical requirements of related application fields.
In one embodiment, a method of landmark image recognition is provided, and the step S140 includes the step S440 of:
and step S440, matching the reference image with the equivalent image sequence according to the gray scale invariant model, and acquiring the equivalent image sequence and the deformation field data of the displacement of the reference image.
And matching the reference image with the equivalent image according to the gray-scale invariant model, acquiring displacement data of each point on the surface of the equivalent image, and generating deformation field data of the displacement of the equivalent image sequence and the reference image.
In one embodiment, a method for identifying a landmark image is provided, and the step S120 includes the step S520:
in step S520, a group of short image sequences with a preset sequence length is selected from the image sequences of the mark points.
Wherein, short image sequences with different sequence lengths can be selected by changing the length of the preset sequence. The longer the short image sequence length, the higher the image quality of the equivalent image obtained equivalently. Therefore, the length of the preset sequence is increased, the image quality of the equivalent image can be improved, the error of the displacement obtained by final calculation can be further reduced, and the precision of the displacement calculation is improved.
According to the method for identifying the image of the mark point, a group of short image sequences are selected according to the preset sequence length in the image sequences of the mark point, the short image sequences are equivalent to equivalent images, the reference image and the equivalent image sequences are subjected to matching analysis according to the gray scale invariant model, the displacement function of the equivalent images is obtained, and the calculation precision is improved by improving the image quality of the equivalent images, so that the identification precision of the image of the mark point is improved.
In one embodiment, a method of landmark image recognition is provided, and step S140 includes step S640:
step 640, obtaining the displacement of the equivalent image of the selected short image sequence relative to the reference image according to the gray scale invariant model; taking the region where the short image sequence is located as a sequence selection region, translating the sequence selection region in the time direction of the image sequence of the mark point by a preset image quantity length, updating the short image sequence, and calculating corresponding displacement until all the image sequences of the mark point are selected; and acquiring deformation field data according to all the calculated displacements.
And matching the reference image with the equivalent image according to the gray-scale invariant model, acquiring displacement data of each point on the surface of the equivalent image, and generating deformation field data of the displacement of the equivalent image sequence and the reference image.
It should be further explained that a first group of short image sequences are equivalent to an equivalent image, displacement is calculated, the region where the short image sequence is located is used as a sequence selection region, the sequence selection region is translated by a preset image number length in the time direction of the image sequence of the mark point, images with the preset image number length are loaded, images with the preset image number length in the first group of short image sequences are removed, the loaded images are arranged behind the images of the previous group of short image sequences to form a second group of short image sequences, the second group of short image sequences are equivalent to an equivalent image, displacement is calculated, and so on until all the images in the image sequence of the mark point are selected, and deformation field data are obtained according to all the calculated displacements.
For example, from time t0 to time t8, the image capturing apparatus acquires 9 consecutive images of the marker point, which constitute an image sequence of the marker point, the sequence length of the preset short image sequence is 3, selects a first group of short image sequences, whose images are P1, P2, and P3, respectively, equates the first group of short image sequences to one equivalent image, calculates the displacement, regards the region where the short image sequence is located as a sequence selection region, shifts the sequence selection region by the preset image number length in the time direction of the image sequence of the marker point, sets the preset image number length to 1, loads one image, removes the previous image P1 in the first group of short image sequences, and arranges the loaded image P4 behind the images of the first group of short image sequences, which constitute a second group of short image sequences, and the images of the second group of short image sequences are P2, P3 and P4, the second group of short image sequences are equivalent to an equivalent image, the displacement is calculated, and the like is performed until all the images in the image sequence of the mark point are selected, and deformation field data are obtained according to all the calculated displacements.
According to the method for identifying the mark point image, the displacement of the equivalent image relative to the reference image is obtained according to the gray scale invariant model, the images with the preset image quantity and length are loaded in the time direction of the image sequence of the mark point, the short image sequence is updated, the corresponding displacement is calculated until all the images in the image sequence of the mark point are selected, deformation field data are obtained according to all the calculated displacements, the displacement of the equivalent image is calculated in real time, the time domain noise of the mark point image is reduced, the problem of low identification precision of the mark point image is solved, the influence of time and space inconsistency of the gray scale of the mark point image on the identification precision of the mark point image is reduced, and high-precision and real-time identification calculation of the mark point image is realized.
In one embodiment, a method for identifying a landmark image is provided, and the step S110 includes the step S710:
step S710, selecting the image subarea containing the mark point, acquiring a plurality of images of the continuously acquired image subareas to form an image sequence of the mark point, and selecting a partial image from the image sequence of the mark point as a reference image.
The method comprises the following steps that the mark points can be positions to be monitored in medical feature positioning, artificial intelligence AI control and engineering structure deformation measurement, a monitoring area is selected as the mark points according to the requirements of practical application, a plurality of mark point images in a time period needing to be monitored are collected through image collection equipment, and the collection number of the mark point images can be determined according to the requirements of practical research and calculation cost; furthermore, an image sub-area containing the mark point is selected as an area for identifying and monitoring the mark point image, and the mark point is ensured to be in the image sub-area in the whole monitoring period.
By the method for identifying the image of the mark point, the image sub-area including the mark point is selected, the images of the plurality of continuously acquired image sub-areas are acquired, and the image sub-area identified by the mark point is accurately positioned on the premise of ensuring that the mark point is always in the image sub-area in the whole deformation process, so that the efficiency of image identification is improved, and the cost of image identification and the cost of calculation are reduced.
In one embodiment, a method for identifying a landmark image is provided, and the step S110 includes the step S810 of:
and step S810, acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points, continuously selecting a plurality of images in the image sequence of the mark points, and equivalently converting the selected plurality of images into a reference image.
It should be noted that the mark points may be positions to be monitored in medical feature positioning, artificial intelligence AI control and engineering structure deformation measurement, a monitoring area is selected as the mark points according to the requirements of practical application, a plurality of mark point images within a time period to be monitored are acquired through image acquisition equipment, and the acquisition number of the mark point images can be determined according to the requirements of practical research and cost calculation.
By the method for identifying the marker point images, the reference images of a plurality of marker points are equivalent to the reference image of one marker point, so that the image quality of the reference image of the marker point is improved, and the identification precision of the marker point image is further improved.
The application also provides the following three specific embodiments, which further explain the method for identifying the mark point image in detail, including embodiment 1, embodiment 2 and embodiment 3:
in embodiment 1, the above method for identifying a landmark image includes the steps of:
step S910, fig. 2 is a schematic diagram of an image sequence for simulating a marker point vibration process according to an embodiment of the present invention, as shown in fig. 2, a reference image is selected, the reference image includes a marker point with a size of 6 × 6pixels, an image sub-region with a size of 15 × 15pixels is selected to ensure that the marker point is always in the image sub-region in the whole deformation process, a sine function y equal to 1.5sinx is applied to the reference image by a simulation method to obtain a deformed image, and a gaussian noise with a standard deviation σ equal to 2(255 gray levels) and a mean value of 0 is introduced into the deformed image to generate a deformed image sequence of the marker point.
In step S920, a deformed image sequence of the mark point with a preset sequence length is selected from the deformed image sequences of the mark points to form a short image sequence of the mark point.
In step S930, the short image sequence is equivalent to an equivalent image, and an equivalent image sequence in which a plurality of equivalent images constitute a marker point is generated.
And step S940, according to the gray scale invariant model, matching analysis is carried out on the reference image and the equivalent image sequence, and deformation field data of the displacement of the equivalent image sequence and the reference image are obtained.
In this embodiment, fig. 3 is a first schematic diagram of displacement-time curves of the mark points measured according to different methods, and as shown in fig. 3, it can be known that displacement (displacement/pixel) -time (time/s) curves of the mark points measured by a gray scale centroid method (gray scale barycentric), a conventional dic (digital computer) method, and a mark point image recognition method provided by the present invention are compared with a theoretical curve (temporal curve), and the displacement fluctuation obtained by using the gray scale centroid method is larger because the gray scale centroid method is more sensitive to the change of the space gray scale of the mark image compared with other methods. It can be seen from the enlarged portion of fig. 3 that the displacement measurement results of the conventional DIC method deviate from the theoretical values at some time. The above-mentioned methods for identifying the marker point images of short image sequences with preset sequence lengths of 3 and 5 are called STS-DIC-3 and STS-DIC-5, respectively, and the measurement results thereof are matched with theoretical values.
Furthermore, the measurement effect can also be quantitatively evaluated by calculating displacement measurement errors of different methods. The systematic error and the random error are calculated by using a formula to obtain the random error and the systematic error of displacement measurement of different analysis methods, fig. 4 is a schematic diagram of the random error and the systematic error of displacement measurement according to different methods, as shown in table 4, according to the result of fig. 4, the method for identifying the mark point image has the smallest measurement error and the largest measurement error of the gray scale gravity center method when the preset sequence length of the selected short image sequence is 5. Compared with the traditional DIC, the measurement error of the marker point image identification method is reduced by two thirds. Through comparison, the calculation cost of the method provided by the invention is increased along with the increase of the preset sequence length of the short image sequence, but the increase of the calculation cost is not a main problem compared with the improvement of the measurement precision of the method provided by the invention. According to the above analysis, although the calculation cost is increased, the identification accuracy of the method for identifying the landmark point image proposed by the present invention is superior to that of the conventional DIC and the gray center of gravity method.
In embodiment 2, the above-described marker point image recognition method includes the steps of:
step S1010, in the uniaxial tension experiment, a polymer test piece in the shape of a dog bone is sprayed by black spray paint, two white mark points were then made at both ends, fig. 5 is a schematic illustration of a uniaxial tensile test with mark points according to an embodiment of the invention, as shown in fig. 5, stretching the sample on a tensile testing machine at a loading rate of 1mm/min, collecting the marked point image on the surface of the sample at a collecting rate of 1 frame/second by using a high-resolution camera, the deformation of the test piece can be regarded as linear elasticity in the initial loading stage, and the deformation image in the camera recording stage is used for acquiring the images of the image subareas where a plurality of continuous mark points are positioned to form an image sequence of the mark points, depending on the current imaging settings, each white marker point occupies approximately 105 x 105pixels, setting the image sub-region size to 121 x 121pixels ensures that the marker points are within the image sub-region throughout the deformation process.
Step S1020, selecting multiple groups of short image sequences of the mark points from the deformed image sequences of the mark points, where the difference between each group of short image sequences is a preset image number length.
And step S1030, respectively equating the multiple groups of short image sequences into multiple equivalent images, and taking the multiple equivalent images as equivalent image sequences.
And step S1040, matching the reference image with the equivalent image sequence according to the gray scale invariant model, and obtaining the equivalent image sequence and the deformation field data of the displacement of the reference image.
In this embodiment, in consideration of the calculation cost, when the method of the present invention is adopted, the length of the selected short image sequence is 3, which is referred to as STS-DIC-3. Fig. 6 is a second schematic diagram of displacement-time curves of the marker points measured according to different methods, as shown in fig. 6. Fig. 7 is a third schematic diagram of the strain-time curves of the marker points measured according to different methods, as shown in fig. 7. According to the displacement-time curve (displacement/time) and the strain-time curve (strain/time) of the mark point, the method provided by the invention has the strongest smoothing effect, the traditional DIC method is second, the robustness of the displacement result measured by using the gray scale gravity center method is worst, and the superiority of the method for identifying the mark point image provided by the invention is verified.
In embodiment 3, the above-mentioned method for identifying a landmark image includes the steps of:
step S1110, fig. 8 is a schematic diagram of an experimental apparatus for truss motion measurement of an expandable antenna according to an embodiment of the present invention, as shown in fig. 8, there are two marker points on a target to be measured, a distance between the two marker points is kept constant during a measurement process, the distance between the two points can be obtained by an image measuring instrument, the distance is used as a reference value for measurement, during the expansion process, a 3D motion measurement system is used to shoot the expansion process of the truss, before starting the measurement, a camera is calibrated to obtain internal parameters and external parameters of the camera, an image sub-region selects 31 × 31pixels, during the measurement process, an image acquisition frequency of the camera is 30 frames/second, and a plurality of continuous images of the two marker points are respectively acquired to form an image sequence of the two marker points.
Step S1120, acquiring a plurality of continuously acquired images of the mark point to form an image sequence of the mark point, selecting a partial image from the image sequence of the mark point as a reference image, and selecting at least one group of short image sequences from the image sequence of the mark point.
In step S1130, the short image sequence is equivalent to an equivalent image, and a plurality of equivalent images are generated to form an equivalent image sequence of the mark point.
Step S1140, according to the gray scale invariant model, obtaining the displacement of the equivalent image relative to the reference image, translating the short image sequence by the preset image number length in the time direction of the image sequence of the mark point, updating the short image sequence, calculating the corresponding displacement until the image sequence of the mark point is all selected, and obtaining the deformation field data according to all the calculated displacements.
In this embodiment, the analysis is performed by using a conventional DIC method, a gray scale barycenter method (grey barycenter) and the method for identifying the mark point image proposed by the present invention, and short image sequences with preset sequence lengths of 3 and 5 are respectively selected during the analysis. First, the image sequences acquired by the two cameras are processed separately using different methods to obtain the 2D motion of the two marker points. It is then used for spatial position correction and three-dimensional reconstruction of the multiple cameras. Finally, the distance between the points A and B is calculated. Fig. 9 is a schematic view of the distance between two marker points measured according to a different method, as shown in fig. 9. According to the measurement result, the result measured by the gray scale gravity center method has the largest fluctuation, the result measured by the method for identifying the mark point image provided by the invention is more reliable and is consistent with a reference value (reference), wherein the result measured by the STS-DIC-5 is closest to the reference value.
The measurement effects of the gray scale gravity center method, the traditional DIC method and the method for identifying the mark point image provided by the invention are compared by means of simulation and experiment. The method for identifying the mark point image provided by the invention is verified, and the identification precision of the mark point image is improved at the cost of slightly increasing the calculation cost.
It should be understood that, although the steps in the flowchart of fig. 1 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 1 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
Corresponding to the manufacturing method of the flexible electronic device, in this embodiment, a device for identifying the image of the mark point is further provided, and the device is used to implement the foregoing embodiment and the preferred embodiment, which have already been described and are not described again. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementations in hardware or a combination of software and hardware are also possible and contemplated.
According to another aspect of the present invention, there is also provided an apparatus for marker point image recognition, fig. 10 is a block diagram showing a configuration of the apparatus for marker point image recognition according to the embodiment of the present invention, as shown in fig. 10, the apparatus including:
the image acquisition module 101 is used for acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points;
an image selecting module 102, configured to select a partial image as a reference image and select at least one group of short image sequences from the image sequences of the mark points;
an image equivalence module 103, configured to equate the short image sequence to an equivalent image;
and the displacement solving module 104 is used for obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model.
In the above device for identifying the mark point image, the image acquisition module 101 is connected with the image selection module 102, the image selection module 102 selects a reference image and a short image sequence from the image sequence acquired by the image acquisition module 101, and the image equivalence module 103 is connected with the image selection module 102 to equate the short image sequence into an equivalent image, thereby reducing the time domain noise of the mark point image, solving the problem of low identification precision of the mark point image, reducing the influence of time and space inconsistency of the gray level of the mark point image on the identification precision of the mark point image, and finally improving the identification precision of the mark point image, so that the identification precision of the mark point image can meet the technical requirements of related application fields.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method of landmark image recognition, the method comprising:
acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points, and selecting partial images from the image sequence of the mark points as reference images;
selecting at least one group of short image sequences from the image sequences of the mark points;
equating the short image sequence to an equivalent image;
and obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model.
2. The method according to claim 1, wherein the selecting at least one group of short image sequences from the image sequences of the marker points comprises:
and selecting a plurality of groups of short image sequences of the mark points from the image sequences of the mark points, wherein the difference between two adjacent groups of short image sequences is a preset image quantity length.
3. The method of claim 2, wherein said equating the sequence of short images to equivalent images comprises:
and respectively equating the plurality of groups of short image sequences into a plurality of equivalent images, and taking the equivalent images as equivalent image sequences.
4. The method of claim 3, wherein obtaining the displacement of the equivalent image relative to the reference image according to a gray-scale invariant model comprises:
and matching the reference image with the equivalent image sequence according to a gray-scale invariant model to obtain the equivalent image sequence and the deformation field data of the displacement of the reference image.
5. The method according to claim 1, wherein the selecting at least one group of short image sequences from the image sequences of the marker points comprises:
and selecting a group of short image sequences in the image sequences of the mark points according to a preset sequence length.
6. The method of claim 5, wherein obtaining the displacement of the equivalent image relative to the reference image according to a gray-scale invariant model comprises:
obtaining the displacement of the equivalent image of the selected short image sequence relative to the reference image according to the gray-scale invariant model;
taking the area where the short image sequence is located as a sequence selection area, translating the sequence selection area by a preset image quantity length in the time direction of the image sequence of the mark point, updating the short image sequence, and calculating corresponding displacement until all the image sequences of the mark point are selected;
and acquiring deformation field data according to all the calculated displacements.
7. The method of claim 1, wherein the model that is invariant in terms of gray scale comprises:
wherein X is the displacement value of the mark point, U is the image displacement of the mark point, α is the related parameter, delta t is the time interval between two images, FiRepresenting the gray value of the reference image, GiThe gray value of the deformed image is shown, and m represents m frames of images before and after the ith frame.
8. The method of claim 1, wherein said acquiring a plurality of successively acquired images of landmark points comprises:
and selecting an image subregion containing the mark point to obtain a plurality of continuously acquired images of the image subregion.
9. The method according to claim 1, wherein the selecting a partial image as a reference image in the image sequence of the marker point comprises:
continuously selecting a plurality of images in the image sequence of the mark point;
and equivalently converting the selected multiple images into a reference image.
10. An apparatus for landmark image recognition, the apparatus comprising:
the image acquisition module is used for acquiring a plurality of continuously acquired images of the mark points to form an image sequence of the mark points;
the image selection module is used for selecting partial images from the image sequences of the mark points as reference images and selecting at least one group of short image sequences;
the image equivalence module is used for enabling the short image sequence to be equivalent to an equivalent image;
and the displacement solving module is used for obtaining the displacement of the equivalent image relative to the reference image according to the gray-scale invariant model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010042272.3A CN111274897A (en) | 2020-01-15 | 2020-01-15 | Method and device for identifying mark point image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010042272.3A CN111274897A (en) | 2020-01-15 | 2020-01-15 | Method and device for identifying mark point image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111274897A true CN111274897A (en) | 2020-06-12 |
Family
ID=71003237
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010042272.3A Pending CN111274897A (en) | 2020-01-15 | 2020-01-15 | Method and device for identifying mark point image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111274897A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114993243A (en) * | 2022-08-04 | 2022-09-02 | 深圳粤讯通信科技有限公司 | Antenna attitude monitoring and early warning system based on Internet of things |
-
2020
- 2020-01-15 CN CN202010042272.3A patent/CN111274897A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114993243A (en) * | 2022-08-04 | 2022-09-02 | 深圳粤讯通信科技有限公司 | Antenna attitude monitoring and early warning system based on Internet of things |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107492127B (en) | Light field camera parameter calibration method and device, storage medium and computer equipment | |
| Steinbrücker et al. | Large displacement optical flow computation withoutwarping | |
| CN106875443B (en) | The whole pixel search method and device of 3-dimensional digital speckle based on grayscale restraint | |
| CN103793915B (en) | Inexpensive unmarked registration arrangement and method for registering in neurosurgery navigation | |
| CN110610486B (en) | Monocular image depth estimation method and device | |
| CN106023303A (en) | Method for improving three-dimensional reconstruction point-clout density on the basis of contour validity | |
| CN114663509B (en) | Self-supervision monocular vision odometer method guided by key point thermodynamic diagram | |
| CN109373912A (en) | A kind of non-contact six-freedom displacement measurement method based on binocular vision | |
| EP4066162A1 (en) | System and method for correspondence map determination | |
| CN111274897A (en) | Method and device for identifying mark point image | |
| KR20220093492A (en) | System and method for establishing structural exterior map using image stitching | |
| CN114820563A (en) | Industrial component size estimation method and system based on multi-view stereo vision | |
| CN109087279A (en) | A kind of object deflection fast acquiring method based on digital picture diffraction | |
| JP4102386B2 (en) | 3D information restoration device | |
| JP5267100B2 (en) | Motion estimation apparatus and program | |
| CN116664531A (en) | Deep learning-based large deformation measurement method and system | |
| CN116109778A (en) | Face three-dimensional reconstruction method based on deep learning, computer equipment and medium | |
| CN114485417B (en) | Structural vibration displacement identification method and system | |
| CN105430397A (en) | 3D (three-dimensional) image experience quality prediction method and apparatus | |
| CN113537351B (en) | Remote sensing image coordinate matching method for mobile equipment shooting | |
| CN114255200A (en) | Myocardial stress analysis method and device based on real-time film imaging | |
| CN103208013A (en) | Photo source identification method based on image noise analysis | |
| CN114170321A (en) | Camera self-calibration method and system based on distance measurement | |
| CN113610906A (en) | Fusion image guidance-based multi-parallax image sequence registration method | |
| CN115222640A (en) | Magnetic resonance image correction method, magnetic resonance image correction device, computer equipment and storage medium |
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 |