CN114723967B - Disparity map optimization method, face recognition device, equipment and storage medium - Google Patents
Disparity map optimization method, face recognition device, equipment and storage medium Download PDFInfo
- Publication number
- CN114723967B CN114723967B CN202210234171.5A CN202210234171A CN114723967B CN 114723967 B CN114723967 B CN 114723967B CN 202210234171 A CN202210234171 A CN 202210234171A CN 114723967 B CN114723967 B CN 114723967B
- Authority
- CN
- China
- Prior art keywords
- disparity
- matching
- value
- pixel point
- confidence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Abstract
The embodiment of the application relates to the field of image processing, and discloses a disparity map optimization method, a face recognition device, equipment and a storage medium. The disparity map optimization method comprises the following steps: aiming at each pixel point in the initial disparity map, calculating the matching confidence of the pixel point according to the matching cost of the initial disparity value of the pixel point and the matching costs of a plurality of disparity values adjacent to the initial disparity value; generating a confidence map according to the matching confidence of each pixel point, and performing connected domain detection on the confidence map to determine a plurality of connected domains; and determining an unreliable connected domain according to the area of the connected domain and the matching confidence coefficient mean value of the connected domain, and removing the parallax value corresponding to the pixel point in the unreliable connected domain from the initial parallax image to obtain the optimized parallax image. Whether the disparity value is reliable or not is evaluated through the area of the connected domain and the confidence coefficient mean value of the connected domain, and therefore a more accurate disparity map is obtained quickly.
Description
Technical Field
The embodiment of the application relates to the field of face recognition, in particular to a disparity map optimization method, a face recognition device, equipment and a storage medium.
Background
A structured light camera is a camera that obtains depth data by emitting an active infrared light source, and is widely used in the field of three-dimensional face recognition, such as: a payment scenario, a door lock scenario, and a rail crossing scenario. The depth reconstruction of the structured light camera utilizes a triangular ranging method to calculate the distance from a measured object to the camera, a disparity map is required to be obtained through a stereo matching algorithm in the process, and the stereo matching algorithm mainly comprises four steps: matching cost calculation, cost aggregation, parallax calculation and parallax optimization.
However, in the process of obtaining the disparity map by performing matching calculation on the object speckle pattern and the reference speckle pattern, mismatching is easy to occur, and further, the occurrence of wrong disparity is caused. Therefore, in order to improve the reliability of the disparity map, a left-right consistency method is usually adopted to eliminate false disparity, but in the process of acquiring the optimized disparity map in this way, stereo matching needs to be performed twice, so that the calculation amount is greatly increased, and the efficiency is low.
Disclosure of Invention
The embodiment of the application aims to provide a disparity map optimization method, a face recognition device, equipment and a storage medium, and whether the disparity value is reliable or not is evaluated through the area of a connected domain and the confidence coefficient average value of the connected domain, so that a more accurate disparity map can be rapidly acquired.
In order to solve the above technical problem, an embodiment of the present application provides a disparity map optimization method, including: aiming at each pixel point in the initial disparity map, calculating the matching confidence of the pixel point according to the matching cost of the initial disparity value of the pixel point and the matching costs of a plurality of disparity values adjacent to the initial disparity value; generating a confidence map according to the matching confidence of each pixel point, and performing connected domain detection on the confidence map to determine a plurality of connected domains; and determining an unreliable connected domain according to the area of the connected domain and the matching confidence coefficient mean value of the connected domain, and removing the parallax value corresponding to the pixel point in the unreliable connected domain from the initial parallax image to obtain the optimized parallax image.
In order to solve the above technical problem, an embodiment of the present application provides a face recognition method, including: optimizing an initial face disparity map of a face to be recognized according to the disparity map optimization method in the embodiment to obtain an optimized face disparity map; and generating a face depth map according to the optimized face disparity map, and matching the face depth map with a preset face depth map in a preset face database to obtain a face recognition result.
An embodiment of the present application further provides a disparity map optimization apparatus, including:
the confidence coefficient calculation module is used for calculating the matching confidence coefficient of each pixel point in the initial disparity map according to the matching cost of the initial disparity value of the pixel point and the matching costs of a plurality of disparity values adjacent to the initial disparity value; generating a confidence map according to the matching confidence of each pixel point, and performing connected domain detection on the confidence map to determine a plurality of connected domains;
and the disparity map optimization module is used for determining an unreliable connected domain according to the area of the connected domain and the matching confidence coefficient mean value of the connected domain, and removing disparity values corresponding to pixel points in the unreliable connected domain from the initial disparity map to obtain an optimized disparity map.
An embodiment of the present application also provides an electronic device, including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executable by the at least one processor to enable the at least one processor to perform the disparity map optimization method as mentioned in the above embodiments, or to perform the face recognition method as mentioned in the above embodiments.
Embodiments of the present application also provide a computer-readable storage medium storing a computer program, where the computer program is executed by a processor to implement the disparity map optimization method mentioned in the above embodiments, or to be capable of executing the face recognition method mentioned in the above embodiments.
According to the disparity map optimization method provided by the embodiment of the application, the matching confidence of each pixel point is calculated according to the matching cost of the initial disparity value of each pixel point in the initial disparity map and the matching costs of a plurality of disparity values adjacent to the initial disparity value, a confidence map is generated according to the matching confidence, whether the disparity value of each pixel point is reliable or not is evaluated according to the area of each connected domain in the confidence map and the mean value of the matching confidence of the connected domain, on one hand, the confidence map is introduced to replace the stereo matching process in the left-right consistency method, the problem of large calculation amount caused by pixel-by-pixel matching is avoided, the confidence can be calculated only according to the matching costs, and the calculation amount is greatly reduced; on the other hand, the reliability map is subjected to connected domain detection to determine a plurality of connected domains, the unreliable parallax is determined according to two conditions of the area of the connected domains and the mean value of the matching confidence of the connected domains, the unreliable parallax is prevented from being determined by mistake only depending on the matching confidence or the area of the connected domains, and the parallax map is more accurate and reliable.
In addition, the method for optimizing a disparity map, which determines an unreliable region in the confidence map according to the area of the connected component and the mean value of the confidence coefficients of matching of the connected component, according to the embodiment of the present application, includes: and when the area of the connected domain is smaller than a preset area threshold value and the mean value of the matching confidence coefficients of the connected domain is smaller than a preset confidence coefficient threshold value, determining that the connected domain is an unreliable region. When the area of the connected domain and the mean value of the matching confidence of the connected domain both meet the conditions, the parallax value corresponding to the pixel point in the connected domain is considered to be unreliable parallax, namely whether the parallax value of the pixel point in the region is reliable is macroscopically measured from the angle of one region, and whether the parallax value is reliable is not judged one by one from the angle of one pixel point, so that the calculated amount is greatly reduced, the optimization efficiency is improved, and the misidentification of mismatching pixel points during the judgment of a single condition can be avoided.
In addition, the disparity map optimization method provided in the embodiment of the present application, after obtaining the optimized disparity map, further includes: and aiming at each pixel point in the optimized disparity map, performing sub-pixel difference according to the matching cost of the disparity value of the pixel point and the matching cost of a plurality of disparity values adjacent to the disparity value to obtain the disparity map at the sub-pixel level. After the unreliable disparity is eliminated, sub-pixel interpolation is carried out on the disparity map according to a plurality of matching costs, so that the accuracy of the obtained disparity map is higher.
Drawings
One or more embodiments are illustrated by way of example in the accompanying drawings which correspond to and are not to be construed as limiting the embodiments, in which elements having the same reference numeral designations represent like elements throughout, and in which the drawings are not to be construed as limiting in scale unless otherwise specified.
Fig. 1 is a flowchart of a disparity map optimization method provided in an embodiment of the present application;
fig. 2 is a schematic structural diagram of a disparity map optimizing apparatus provided in an embodiment of the present application;
fig. 3 is a schematic structural diagram of an electronic device provided in an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following describes each embodiment of the present application in detail with reference to the accompanying drawings. However, it will be appreciated by those of ordinary skill in the art that numerous technical details are set forth in various embodiments of the present application in order to provide a better understanding of the present application. However, the technical solution claimed in the present application can be implemented without these technical details and various changes and modifications based on the following embodiments.
The following describes details of implementation of the disparity map optimization according to the present embodiment. The following disclosure provides implementation details for the purpose of facilitating understanding, and is not necessary to practice the present solution.
The parallax image optimization method provided by the application mainly carries out further optimization on the parallax image obtained after parallax matching calculation, eliminates unreliable parallax, improves parallax precision, and enables the parallax image to be more reliable and accurate. The currently common method for eliminating unreliable parallax is a left-right consistency method, and the specific process is as follows: after obtaining the parallax images (left image: object speckle pattern, right image: reference speckle pattern), interchanging the left image and the right image, namely, the left image is the reference speckle pattern, the right image is the object speckle pattern,after one-time stereo matching, a new disparity map is obtained, and for each point p in the disparity map, the homonymy point p is determined in the new disparity map 1 And when the difference value of the parallax values of the same-name points meets a preset condition, determining the parallax value of the p point in the parallax map as the unreliable parallax. Therefore, when the left-right consistency is determined to be unreliable parallax, the left-right consistency needs to be determined according to the two parallax images, namely, a stereo matching process needs to be performed again, the calculation amount of the stereo matching process is large, and the efficiency is low.
The embodiment of the present application relates to a disparity map optimization method, as shown in fig. 1, including:
Specifically, the initial disparity map may perform disparity matching calculation on the object speckle map and the reference speckle map through a preset image matching algorithm, obtain a matching cost curve (the horizontal axis represents disparity, and the vertical axis represents matching cost) of each pixel point, and use a disparity value corresponding to a minimum matching cost value in the matching cost curve as the initial disparity value. And the initial parallax value of each pixel point in the initial parallax image is the parallax value corresponding to the minimum matching cost value of the pixel point. It should be noted that the preset image matching algorithm may be one or more combinations of image matching algorithms such as a mean absolute difference algorithm (MAD), a sum of absolute differences algorithm (SAD), a sum of squared errors algorithm (SSD), a mean sum of error algorithm (MSD), a normalized product correlation algorithm (NCC), and a Sequential Similarity Detection Algorithm (SSDA), which are not limited herein.
The matching cost of the plurality of disparity values adjacent to the initial disparity value may be the matching cost of the plurality of disparity values adjacent to the right side of the initial disparity value, the matching cost of the plurality of disparity values adjacent to the left side of the initial disparity value, or the matching cost of a plurality of disparity values selected on the left side and the right side of the initial disparity value. Such as: for an initial parallax value D of a certain pixel point, calculating the matching confidence of the pixel point according to the matching cost of the initial parallax value D, the matching cost of D +1 and the matching cost of D + 2; the matching confidence of the pixel point can also be calculated according to the matching cost of the initial parallax value D, the matching cost of D-1 and the matching cost of D-2; and calculating the matching confidence of the pixel point according to the matching cost of the initial parallax value D, the matching cost of D-1 and the matching cost of D + 1. It should be noted that, in general, the more obvious the valley value of the matching cost curve is, the higher the matching confidence is, i.e. the more reliable the disparity value is.
In one embodiment, the matching confidence for each pixel point is calculated by the following formula:
wherein cost (x, y, D) is a matching cost of the initial disparity value, and cost (x, y, D ± k) is a matching cost of a plurality of disparity values adjacent to the initial disparity value.
And 102, generating a confidence map according to the matching confidence of each pixel point, carrying out connected domain detection on the confidence map, and determining a plurality of connected domains.
Specifically, the detecting the connected components of the confidence map to determine a plurality of connected components includes: taking each pixel point in the confidence map as a seed pixel point, and calculating the difference value between each pixel point and the seed pixel point in the field of the seed pixel point; and determining a plurality of connected domains according to the difference value and a preset connected domain threshold value.
And performing connected domain detection on the reliability map, taking eight-connected domain detection as an example, detecting the difference value between 8 pixel points in the field of each pixel point in the reliability map and the pixel point, namely detecting the difference value between the 8 pixel points of the upper, lower, left, right, upper left, lower left, upper right and lower right of a certain pixel point and the pixel point, if the difference value is smaller than a preset connected domain threshold value, considering that two pixel points corresponding to the difference value are the same connected domain, and if the difference value is greater than or equal to the preset connected domain threshold value, considering that two pixel points corresponding to the difference value do not belong to the same connected domain, thus determining a plurality of connected domains. Of course, the connected component detection may be other detection methods such as four-connected component detection.
And 103, determining an unreliable connected domain according to the area of the connected domain and the matching confidence coefficient mean value of the connected domain, and removing the parallax value corresponding to the pixel point in the unreliable connected domain from the initial parallax image to obtain an optimized parallax image.
In this embodiment, when the area of the connected domain is smaller than a preset area threshold and the mean value of the matching confidence of the connected domain is smaller than a preset confidence threshold, it is determined that the connected domain is an unreliable region.
It should be noted that the connected domain refers to a region formed by pixel points whose pixel values are relatively close and whose positions are adjacent to each other. The connected domain is calculated through the comparison of the confidence maps, that is, the matching confidence values of the pixels belonging to the same connected domain are relatively close, and when the area of the connected domain is too small (far smaller than a preset area threshold), for example, only 1 pixel in a certain connected domain indicates that no matching confidence value which is relatively close to the matching confidence value of the pixel exists at the position adjacent to the pixel, so that the parallax value corresponding to the pixel in the connected domain can be regarded as unreliable parallax.
And the matching confidence coefficient is calculated according to the matching cost of the initial parallax value and the matching costs of a plurality of parallax values adjacent to the initial parallax value, and the smaller the matching confidence coefficient is, the closer the matching cost of the initial parallax value is to the matching costs of the plurality of parallax values adjacent to the initial parallax value, that is, the less reliable the initial parallax value is (the reliable parallax value should be the parallax value corresponding to the point with the minimum matching cost value, and there is only one point with the minimum matching cost value).
It is worth mentioning that, the method calculates the area of each connected domain and the mean value of the matching confidence of all the pixels in the connected domain, when both the area of the connected domain and the mean value of the matching confidence meet the condition, the disparity value corresponding to the pixels in the connected domain is considered to be unreliable disparity, it can be understood that, when the mean value of the matching confidence of the connected domain meets the condition, the matching confidence of all the pixels in the connected domain may be smaller than the preset confidence threshold, or there may be both pixels in the connected domain that are larger than the confidence threshold and pixels that are smaller than the confidence threshold, that is, the method macroscopically measures whether the disparity value of the pixels in the connected domain is reliable from the angle of one region, rather than judging whether the disparity value is reliable from the angle of one pixel one by one, thereby not only greatly reducing the calculation amount and improving the optimization efficiency, but also avoiding the false determination of the mismatched pixels when a single condition is judged.
In addition, the confidence threshold is set in relation to the magnitude of the confidence value, i.e., in relation to the magnitude of the matching cost value, which is often related to the size of the matching window. In addition, when the unreliable disparity is removed from the initial disparity map, the unreliable disparity can be directly assigned to 0.
In addition, after obtaining the optimized disparity map, the method further includes: and aiming at each pixel point in the optimized disparity map, performing sub-pixel difference according to the matching cost of the disparity value of the pixel point and the matching cost of a plurality of disparity values adjacent to the disparity value to obtain the disparity map at the sub-pixel level. Specifically, when performing sub-pixel interpolation on the optimized disparity map, the matching cost of a plurality of disparity values adjacent to each pixel disparity value may be selected for calculation, for example: for the optimized disparity value d, sub-pixel interpolation can be performed according to the matching cost corresponding to d, the matching cost corresponding to d +1 and the matching cost corresponding to d-1, and sub-pixel interpolation can also be performed according to the matching cost corresponding to d, the matching cost corresponding to d +2 and the matching cost corresponding to d-2.
Specifically, sub-pixel interpolation is performed through the following formula to obtain a disparity map at a sub-pixel level:
wherein, D ' is a disparity value of each pixel in the optimized disparity map, cost (x, y, D ') is a matching cost of the pixel (x, y) when the disparity value is D ', cost (x, y, D ' -1) is a matching cost of the pixel (x, y) when the disparity value is D ' -1, and cost (x, y, D ' + 1) is a matching cost of the pixel (x, y) when the disparity value is D ' + 1.
According to the disparity map optimization method provided by the embodiment of the application, the matching confidence coefficient of each pixel point is calculated according to the matching cost of the initial disparity value of each pixel point in the initial disparity map and the matching costs of a plurality of disparity values adjacent to the initial disparity value, a confidence map is generated according to the matching confidence coefficient, and whether the disparity value of each pixel point is reliable or not is evaluated according to the area of each connected domain in the confidence map and the mean value of the matching confidence coefficient of the connected domain; on the other hand, the reliability map is subjected to connected domain detection to determine a plurality of connected domains, the unreliable parallax is determined according to two conditions of the area of the connected domains and the mean value of the matching confidence of the connected domains, the unreliable parallax is prevented from being determined by mistake only depending on the matching confidence or the area of the connected domains, and the parallax map is more accurate and reliable.
The embodiment of the application also relates to a face recognition method, which comprises the following steps: optimizing the initial face disparity map of the face to be recognized according to the disparity map optimization method provided by the above embodiment, and acquiring the optimized face disparity map; and generating a face depth map according to the optimized face disparity map, and matching the face depth map with a preset face depth map in a preset face database to obtain a face recognition result.
According to the face recognition method, the face depth map is generated according to the optimized face disparity map, and face recognition is carried out through the face depth map, so that the face false recognition caused by the generation of wrong face depth map due to the fact that false matching is caused in the process of generating the initial face disparity map can be avoided.
The steps of the above methods are divided for clarity, and the implementation may be combined into one step or split some steps, and the steps are divided into multiple steps, so long as the same logical relationship is included, which are within the scope of the present patent; it is within the scope of the patent to add insignificant modifications to the algorithms or processes or to introduce insignificant design changes to the core design without changing the algorithms or processes.
Embodiments of the present application relate to a disparity map optimizing apparatus, as shown in fig. 2, including:
a confidence calculation module 201, configured to calculate, for each pixel point in an initial disparity map, a matching confidence of the pixel point according to a matching cost of an initial disparity value of the pixel point and matching costs of a plurality of disparity values adjacent to the initial disparity value; generating a confidence map according to the matching confidence of each pixel point, and performing connected domain detection on the confidence map to determine a plurality of connected domains;
and the disparity map optimization module 202 is configured to determine an unreliable connected domain according to the area of the connected domain and the matching confidence mean of the connected domain, and remove disparity values corresponding to pixel points in the unreliable connected domain from the initial disparity map to obtain an optimized disparity map.
It will be appreciated that this embodiment is an apparatus embodiment corresponding to the method embodiment described above, and that this embodiment can be implemented in cooperation with the above embodiment. The related technical details mentioned in the above embodiments are still valid in this embodiment, and are not described herein again in order to reduce repetition. Accordingly, the related technical details mentioned in the present embodiment can also be applied to the above embodiments.
It should be noted that, all the modules involved in this embodiment are logic modules, and in practical application, one logic unit may be one physical unit, may also be a part of one physical unit, and may also be implemented by a combination of multiple physical units. In addition, in order to highlight the innovative part of the present invention, a unit which is not so closely related to solve the technical problem proposed by the present invention is not introduced in the present embodiment, but this does not indicate that there is no other unit in the present embodiment.
Embodiments of the present application relate to an electronic device, as shown in fig. 3, including:
at least one processor 301; and a memory 302 communicatively coupled to the at least one processor 301; the memory 302 stores instructions executable by the at least one processor 301, and the instructions are executed by the at least one processor 301, so that the at least one processor 301 can perform the disparity map optimization as mentioned in the above embodiments, or can perform the face recognition method as described in the above embodiments.
The electronic device includes: one or more processors 301 and a memory 302, with one processor 301 being illustrated in fig. 3. The processor 301 and the memory 302 may be connected by a bus or other means, and fig. 3 illustrates the connection by a bus as an example. The memory 302 is a non-volatile computer-readable storage medium, which can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the algorithms corresponding to the processing strategies in the strategy space in the embodiment of the present application, in the memory 302. The processor 301 executes various functional applications and data processing of the device, i.e., implements the above-described disparity map optimization method or face recognition method, by running a non-volatile software program, instructions, and modules stored in the memory 302.
The memory 302 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store a list of options, etc. Further, the memory 302 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some embodiments, memory 302 may optionally include memory located remotely from processor 301, which may be connected to an external device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
One or more modules are stored in the memory 302, and when executed by the one or more processors 301, perform the disparity map optimization method in any of the above embodiments, or can perform the face recognition method as described in the above embodiments.
The product can execute the method provided by the embodiment of the application, has corresponding functional modules and beneficial effects of the execution method, and can refer to the method provided by the embodiment of the application without detailed technical details in the embodiment.
Embodiments of the present application relate to a computer-readable storage medium storing a computer program. The computer program realizes the above-described method embodiments when executed by a processor.
That is, as can be understood by those skilled in the art, all or part of the steps in the method according to the above embodiments may be implemented by a program instructing relevant hardware, where the program is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps in the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
It will be understood by those of ordinary skill in the art that the foregoing embodiments are specific examples for carrying out the present application, and that various changes in form and details may be made therein without departing from the spirit and scope of the present application in practice.
Claims (10)
1. A disparity map optimization method, comprising:
aiming at each pixel point in the initial disparity map, calculating the matching confidence of the pixel point according to the matching cost of the initial disparity value of the pixel point and the matching costs of a plurality of disparity values adjacent to the initial disparity value;
generating a confidence map according to the matching confidence of each pixel point, and performing connected domain detection on the confidence map to determine a plurality of connected domains;
and determining an unreliable connected domain according to the area of the connected domain and the matching confidence coefficient mean value of the connected domain, and removing the parallax value corresponding to the pixel point in the unreliable connected domain from the initial parallax image to obtain the optimized parallax image.
2. The method for optimizing a disparity map according to claim 1, wherein the determining unreliable regions in the confidence map according to the areas of the connected components and the mean of the confidence levels of matching of the connected components comprises:
and when the area of the connected domain is smaller than a preset area threshold value and the mean value of the matching confidence coefficients of the connected domain is smaller than a preset confidence coefficient threshold value, determining that the connected domain is an unreliable region.
3. The disparity map optimization method according to claim 1 or 2, wherein the matching confidence of each pixel point is calculated by the following formula:
wherein cost (x, y, D) is a matching cost of the initial disparity value, and cost (x, y, D +/-k) is a matching cost of a plurality of disparity values adjacent to the initial disparity value; and D is the initial parallax value of the pixel point.
4. The disparity map optimization method according to claim 1, wherein the performing connected component detection on the confidence map to determine a plurality of connected components comprises:
taking each pixel point in the confidence map as a seed pixel point, and calculating the difference value between each pixel point in the neighborhood of the seed pixel point and the seed pixel point;
and determining a plurality of connected domains according to the difference value and a preset connected domain threshold value.
5. The method of claim 1, wherein after obtaining the optimized disparity map, the method further comprises:
and performing sub-pixel difference according to the matching cost of the parallax value of each pixel point in the optimized parallax image and the matching cost of a plurality of parallax values adjacent to the parallax value of each pixel point in the optimized parallax image to obtain the parallax image at the sub-pixel level.
6. The method of claim 5, wherein the sub-pixel interpolation is performed according to the following formula to obtain the sub-pixel level disparity map:
wherein D is ’ The disparity value of each pixel point in the optimized disparity map is cost (x, y, D) ’ ) The pixel point (x, y) has a parallax value of D ’ Matching cost of time, cost (x, y, D) ’ -1) as pixel point (x, y) at disparity value D ’ Matching cost at-1, cost (x, y, D) ’ + 1) is the pixel (x, y) with parallax value D ’ Matching cost at + 1.
7. A face recognition method, comprising:
optimizing an initial face disparity map of a face to be recognized according to the disparity map optimization method of any one of claims 1 to 6 to obtain an optimized face disparity map;
and generating a face depth map according to the optimized face disparity map, and matching the face depth map with a preset face depth map in a preset face database to obtain a face recognition result.
8. A disparity map optimization apparatus, comprising:
the confidence coefficient calculation module is used for calculating the matching confidence coefficient of each pixel point in the initial disparity map according to the matching cost of the initial disparity value of the pixel point and the matching costs of a plurality of disparity values adjacent to the initial disparity value; generating a confidence map according to the matching confidence of each pixel point, and performing connected domain detection on the confidence map to determine a plurality of connected domains;
and the disparity map optimization module is used for determining an unreliable connected domain according to the area of the connected domain and the matching confidence coefficient mean value of the connected domain, and removing disparity values corresponding to pixel points in the unreliable connected domain from the initial disparity map to obtain an optimized disparity map.
9. An electronic device, comprising:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the disparity map optimization method according to any one of claims 1 to 6, or to perform the face recognition method according to claim 7.
10. A computer-readable storage medium, storing a computer program, wherein the computer program, when executed by a processor, implements the disparity map optimization method of any one of claims 1 to 6, or is capable of performing the face recognition method of claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210234171.5A CN114723967B (en) | 2022-03-10 | 2022-03-10 | Disparity map optimization method, face recognition device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210234171.5A CN114723967B (en) | 2022-03-10 | 2022-03-10 | Disparity map optimization method, face recognition device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114723967A CN114723967A (en) | 2022-07-08 |
| CN114723967B true CN114723967B (en) | 2023-01-31 |
Family
ID=82237981
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210234171.5A Active CN114723967B (en) | 2022-03-10 | 2022-03-10 | Disparity map optimization method, face recognition device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114723967B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226821A (en) * | 2013-04-27 | 2013-07-31 | 山西大学 | Stereo matching method based on disparity map pixel classification correction optimization |
| WO2018127007A1 (en) * | 2017-01-03 | 2018-07-12 | 成都通甲优博科技有限责任公司 | Depth image acquisition method and system |
| CN110135234A (en) * | 2018-02-08 | 2019-08-16 | 弗劳恩霍夫应用研究促进协会 | For determining confidence level/uncertainty measurement concept of parallax measurement |
| CN111383254A (en) * | 2018-12-29 | 2020-07-07 | Tcl集团股份有限公司 | Depth information acquisition method and system and terminal equipment |
| CN111563921A (en) * | 2020-04-17 | 2020-08-21 | 西北工业大学 | Underwater point cloud acquisition method based on binocular camera |
| CN111784756A (en) * | 2020-06-01 | 2020-10-16 | 安徽奇点智能新能源汽车有限公司 | Parallax confidence estimation method and device, electronic equipment and medium |
| CN113850859A (en) * | 2020-06-25 | 2021-12-28 | 英特尔公司 | Methods, systems, articles, and apparatus for enhancing image depth confidence maps |
| WO2022021912A1 (en) * | 2020-07-31 | 2022-02-03 | 南方科技大学 | Low power consumption stereo matching system and method for obtaining depth information |
| CN114066950A (en) * | 2021-10-27 | 2022-02-18 | 北京的卢深视科技有限公司 | Monocular speckle structure optical image matching method, electronic device and storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2722816A3 (en) * | 2012-10-18 | 2017-04-19 | Thomson Licensing | Spatio-temporal confidence maps |
| CN110060283B (en) * | 2019-04-17 | 2020-10-30 | 武汉大学 | Multi-measure semi-global dense matching method |
| CN112784874B (en) * | 2020-12-28 | 2022-07-22 | 深兰人工智能芯片研究院（江苏）有限公司 | Binocular vision stereo matching method and device, electronic equipment and storage medium |
-
2022
- 2022-03-10 CN CN202210234171.5A patent/CN114723967B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226821A (en) * | 2013-04-27 | 2013-07-31 | 山西大学 | Stereo matching method based on disparity map pixel classification correction optimization |
| WO2018127007A1 (en) * | 2017-01-03 | 2018-07-12 | 成都通甲优博科技有限责任公司 | Depth image acquisition method and system |
| CN110135234A (en) * | 2018-02-08 | 2019-08-16 | 弗劳恩霍夫应用研究促进协会 | For determining confidence level/uncertainty measurement concept of parallax measurement |
| CN111383254A (en) * | 2018-12-29 | 2020-07-07 | Tcl集团股份有限公司 | Depth information acquisition method and system and terminal equipment |
| CN111563921A (en) * | 2020-04-17 | 2020-08-21 | 西北工业大学 | Underwater point cloud acquisition method based on binocular camera |
| CN111784756A (en) * | 2020-06-01 | 2020-10-16 | 安徽奇点智能新能源汽车有限公司 | Parallax confidence estimation method and device, electronic equipment and medium |
| CN113850859A (en) * | 2020-06-25 | 2021-12-28 | 英特尔公司 | Methods, systems, articles, and apparatus for enhancing image depth confidence maps |
| WO2022021912A1 (en) * | 2020-07-31 | 2022-02-03 | 南方科技大学 | Low power consumption stereo matching system and method for obtaining depth information |
| CN114066950A (en) * | 2021-10-27 | 2022-02-18 | 北京的卢深视科技有限公司 | Monocular speckle structure optical image matching method, electronic device and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| Learning to predict stereo reliability enforcing local consistency of confidence maps;Matteo Poggi, etal.;《2017 IEEE Conference on Computer Vision and Pattern Recognition》;20171109;全文 * |
| 基于ELAS立体匹配算法的研究与改进;张道德等;《湖北工业大学学报》;20170430;第32卷(第2期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114723967A (en) | 2022-07-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Schilling et al. | Trust your model: Light field depth estimation with inline occlusion handling | |
| CN110310320B (en) | Binocular vision matching cost aggregation optimization method | |
| Yang | Fast stereo matching using reliability-based dynamic programming and consistency constraints | |
| CN107316326B (en) | Edge-based disparity map calculation method and device applied to binocular stereo vision | |
| US10949989B2 (en) | Concept for determining a confidence/uncertainty measure for disparity measurement | |
| CN111681256A (en) | Image edge detection method and device, computer equipment and readable storage medium | |
| US9104904B2 (en) | Disparity estimation method of stereoscopic image | |
| CN116029996A (en) | Stereo matching method and device and electronic equipment | |
| WO2022003740A1 (en) | Method for determining the confidence of a disparity map through a self-adaptive learning of a neural network, and sensor system thereof | |
| CN114219717A (en) | Point cloud registration method and device, electronic equipment and storage medium | |
| CN111739071A (en) | Rapid iterative registration method, medium, terminal and device based on initial value | |
| CN114723967B (en) | Disparity map optimization method, face recognition device, equipment and storage medium | |
| Wang et al. | Improving deep stereo network generalization with geometric priors | |
| CN114331919B (en) | Depth recovery method, electronic device, and storage medium | |
| JP2023021087A (en) | Binocular image matching method, apparatus, device, and storage medium | |
| Veld et al. | A novel confidence measure for disparity maps by pixel-wise cost function analysis | |
| JP7152506B2 (en) | Imaging device | |
| CN108764110B (en) | Recursive false detection verification method, system and equipment based on HOG characteristic pedestrian detector | |
| CN111784756A (en) | Parallax confidence estimation method and device, electronic equipment and medium | |
| Zhong et al. | Displacement-invariant cost computation for efficient stereo matching | |
| CN107948629B (en) | Parallax determination method, device and terminal | |
| CN113965697B (en) | Parallax imaging method based on continuous frame information, electronic device and storage medium | |
| Murarka et al. | A meta-technique for increasing density of local stereo methods through iterative interpolation and warping | |
| Mordohai et al. | Stereo using monocular cues within the tensor voting framework | |
| JP7066580B2 (en) | Image processing 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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20221008 Address after: 230091 room 611-217, R & D center building, China (Hefei) international intelligent voice Industrial Park, 3333 Xiyou Road, high tech Zone, Hefei, Anhui Province Applicant after: Hefei lushenshi Technology Co.,Ltd. Address before: 100083 room 3032, North B, bungalow, building 2, A5 Xueyuan Road, Haidian District, Beijing Applicant before: BEIJING DILUSENSE TECHNOLOGY CO.,LTD. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |