CN114842057A - Distance information complementing method, apparatus, storage medium, and computer program product - Google Patents
Distance information complementing method, apparatus, storage medium, and computer program product Download PDFInfo
- Publication number
- CN114842057A CN114842057A CN202210471068.2A CN202210471068A CN114842057A CN 114842057 A CN114842057 A CN 114842057A CN 202210471068 A CN202210471068 A CN 202210471068A CN 114842057 A CN114842057 A CN 114842057A
- Authority
- CN
- China
- Prior art keywords
- distance
- region
- area
- value
- determining
- 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
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000004590 computer program Methods 0.000 title claims abstract description 21
- 230000008602 contraction Effects 0.000 claims description 78
- 238000003384 imaging method Methods 0.000 claims description 45
- 238000012549 training Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 abstract description 2
- 238000004364 calculation method Methods 0.000 description 19
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000012935 Averaging Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 229910003460 diamond Inorganic materials 0.000 description 2
- 239000010432 diamond Substances 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000001788 irregular Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 206010047531 Visual acuity reduced Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 101150049349 setA gene Proteins 0.000 description 1
- 230000009469 supplementation Effects 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the field of data processing, and provides a distance information completion method, a device, a storage medium and a computer program product, wherein the method comprises the following steps: determining a reference region based on a region to be compensated of the depth image; determining a candidate distance value according to the distance value of the effective distance point in the reference area; and completing the distance value of the distance missing point in the area to be completed based on the candidate distance value. The distance information complementing method provided by the invention improves the distance information complementing efficiency by performing the information complementing of the distance missing point on the specific area in the image, and simultaneously improves the accuracy of performing the distance information complementing on the distance missing point in the area to be complemented by comprehensively considering the effective distance point in the reference area corresponding to the area to be complemented to obtain the candidate distance information.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a distance information completion method, device, storage medium, and computer program product.
Background
With the rapid development of computer technology, artificial intelligence has already met the heat trend of one wave and another wave, and the vision-based intelligent identification technology has advanced the daily life of people. In a conventional target detection and identification process, planar information of an object is determined through a planar image obtained by an RGB camera, and then a 3D camera (a camera or a radar capable of performing range imaging) is used to additionally introduce a distance point to perform overall spatial positioning. However, the 3D camera has defects in hardware or interference of a strong reflective object to an internal photosensitive element, the imaging effect of the 3D camera is often unsatisfactory, and accurate and effective distance point information cannot be obtained in a partial image area, which results in wrong guidance for tasks such as robot grabbing, navigation and ranging.
At present, for the missing distance points, a method of complementing the nearest points around the missing points or a method of complementing global points through a complementing network model is mainly adopted. However, the accuracy of distance information completion is low due to poor vision caused by different distances or angles between the object and the camera by adopting the former method; if the latter method is adopted, the distance information completion efficiency is low due to the fact that a large amount of data training is needed and the global point is completed.
Disclosure of Invention
The present invention has been made to solve at least one of the technical problems occurring in the related art. Therefore, the distance information completion method provided by the invention can be used for completing distance points in a specific area and improving the accuracy and efficiency of distance information completion.
The invention further provides the electronic equipment.
The invention also provides a storage medium.
The invention also proposes a computer program product.
The distance information completion method according to the embodiment of the first aspect of the invention comprises the following steps:
determining a reference region based on a region to be compensated of the depth image;
determining a candidate distance value according to the distance value of the effective distance point in the reference area;
completing the distance value of the distance missing point in the area to be completed based on the candidate distance value;
wherein the reference region is within the depth image; the effective distance point is a point where no distance value loss occurs.
According to the distance information complementing method provided by the embodiment of the invention, the information complementing of the distance missing points is carried out on the specific area in the image, so that the distance information complementing efficiency is improved, and meanwhile, the candidate distance information is obtained by comprehensively considering the effective distance points in the reference area corresponding to the area to be complemented, so that the accuracy of carrying out the distance information complementing on the distance missing points in the area to be complemented is improved.
According to one embodiment of the invention, the reference region comprises a constricted region and an expanded region; wherein the contraction region and the expansion region are both within the depth image, the contraction region is located inside the region to be compensated, and the expansion region is located outside the region to be compensated;
the determining a candidate distance value according to the distance value of the effective distance point in the reference region includes:
determining a first candidate distance value according to the distance value of the effective distance point in the contraction area, and determining a second candidate distance value according to the distance value of the effective distance point in the expansion area;
the distance value completion of the distance missing points in the region to be completed based on the candidate distance values includes:
and performing distance value completion on the distance missing points in the region to be completed based on the first candidate distance value and the second candidate distance value.
According to the distance information completion method provided by the embodiment of the invention, the candidate distance information is obtained by comprehensively considering the effective distance points in the contraction area and the expansion area corresponding to the area to be completed, so that the accuracy of performing distance information completion on the distance missing points in the area to be completed is further improved.
According to an embodiment of the present invention, the distance value completion of the distance missing point in the area to be completed based on the first candidate distance value and the second candidate distance value includes:
determining that a difference value between the first candidate distance value and the second candidate distance value is greater than a preset difference threshold value, and performing distance value completion on the distance missing point in the area to be completed by adopting the first candidate distance value;
and determining that the difference value between the first candidate distance value and the second candidate distance value is not greater than the gap threshold value, and performing distance value completion on the distance missing point in the region to be completed by adopting the average value of the first candidate distance value and the second candidate distance value.
According to the distance information completion method provided by the embodiment of the invention, the distance value completion is carried out on the distance missing points in the region to be completed according to different conditions according to the difference value between the first candidate distance value and the second candidate distance value, so that the accuracy of distance information completion is further improved.
According to one embodiment of the present invention, the region to be compensated is a polygon;
the determining of the reference region based on the region to be compensated of the depth image includes:
determining an extension point on at least one edge line of the region to be compensated, determining a plurality of vertexes of the contraction region based on each extension point and a preset adduction distance, and obtaining the contraction region according to the vertexes;
and determining an expansion area central point corresponding to each extension point based on each extension point and a preset outward expansion distance, and obtaining the expansion area according to the expansion area central point.
According to the distance information complementing method provided by the embodiment of the invention, the polygon is adopted to mark the area to be complemented, and one or more contraction areas and one or more expansion areas are determined based on the side line of the polygonal area to be complemented and the preset internal contraction distance or external expansion distance, so that a plurality of relevant areas near the area to be complemented can be uniformly obtained, and the first candidate distance value and the second candidate distance value are comprehensively calculated according to the distance effective values of the relevant areas, thereby further improving the accuracy of distance information complementing.
According to an embodiment of the present invention, after determining the reference region based on the depth image to-be-compensated region, before determining the first candidate distance value according to the distance value of the effective distance point in the shrinkage region and determining the second candidate distance value according to the distance value of the effective distance point in the expansion region, the method further includes:
determining that the number of effective distance points in the contraction area is less than a preset first threshold value, and re-determining the contraction area by adjusting the adduction distance;
and determining that the number of the effective distance points in the expansion area is less than a preset second threshold value, and re-determining the expansion area by adjusting the flaring distance.
According to the distance information complementing method of the embodiment of the invention, when the effective distance points in the contraction area or the expansion area are determined to be less, the distance information of the relevant area is considered to be insufficient for calculating the first candidate distance value or the second candidate distance value, so that the contraction area or the expansion area is re-determined by adjusting the degree of adduction or abduction, the first candidate distance value and the second candidate distance value of the area to be complemented are calculated by acquiring enough distance information, and the accuracy of distance information complementing is further improved.
According to an embodiment of the present invention, determining a first candidate distance value based on the distance values of the effective distance points within the shrinkage region and determining a second candidate distance value based on the distance values of the effective distance points within the expansion region comprises:
determining a first distance average of the shrinkage region, and determining the first candidate distance value according to the first distance average;
determining a second distance average for the expanded region and determining the second candidate distance value based on the second distance average;
the first distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the contraction area; and the second distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the expansion area.
According to the distance information completion method provided by the embodiment of the invention, the average value of the effective distance values of the relevant areas is used as the basis for calculating the candidate distance values, wherein for a plurality of expansion areas or a plurality of contraction areas, the average value of the effective distance values of each expansion area or each contraction area is respectively obtained, and then the corresponding candidate distance values are calculated according to the average values of the distances of the plurality of expansion areas or the plurality of contraction areas, so that the distance point information of the relevant areas around the area to be completed can be uniformly obtained, and the universality and the accuracy of distance information completion are further improved.
According to an embodiment of the present invention, before determining the reference region based on the depth image to-be-compensated region, the method further includes:
determining the region to be compensated from the depth image based on two-dimensional coordinate information of a target identification region of a color imaging image;
wherein the coordinate system of the color imaging graph and the coordinate system of the depth image are registered with each other in advance; the target identification area is an area where an object to be detected is located.
According to the distance information complementing method provided by the embodiment of the invention, the region to be complemented does not need to be directly obtained through the depth image, but the region to be complemented in the depth image is determined through the target identification region of the adaptive color imaging image, and the region of the target object can be identified more quickly and accurately compared with the depth image through the color imaging image, so that the efficiency and the accuracy of distance information complementing are further improved on the whole.
According to an embodiment of the present invention, before determining the region to be compensated from the depth image based on the two-dimensional coordinate information of the target identification region of the color imaging map, the method further includes:
inputting the color imaging graph into a target recognition model to obtain a color imaging graph which is output by the target recognition model and marked with a target recognition area;
the target recognition model is obtained by training based on a sample color imaging graph and a target recognition area result corresponding to the sample color imaging graph.
According to the distance information completion method provided by the embodiment of the invention, the trained target recognition model is adopted to recognize the target object region in the color imaging image, so that the accuracy of selecting the region to be completed by the frame can be further improved, and the accuracy of distance information completion is further improved.
An electronic device according to an embodiment of the second aspect of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the distance information complementing method as described in any one of the above.
A non-transitory computer-readable storage medium according to an embodiment of the third aspect of the present invention, on which a computer program is stored, which, when executed by a processor, implements the distance information complementing method as any one of the above.
The computer program product according to the fourth aspect of the present invention includes a computer program that, when executed by a processor, implements the distance information complementing method as described in any one of the above.
One or more technical solutions in the embodiments of the present invention have at least one of the following technical effects:
the distance information completion efficiency is improved by performing information completion on the distance missing points in the specific area in the image, and meanwhile, the distance information completion accuracy of the distance missing points in the area to be completed is improved by comprehensively considering the effective distance points in the reference area corresponding to the area to be completed to obtain the candidate distance information.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the related arts, the drawings used in the description of the embodiments or the related arts will be briefly introduced below, it is obvious that the drawings in the description below are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without creative efforts.
Fig. 1 is a schematic flow chart of a distance information completion method according to an embodiment of the present invention;
fig. 2 is a schematic overall flow chart of a distance information completion method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a region solution provided by an embodiment of the invention;
fig. 4 is a schematic structural diagram of a distance information completing apparatus according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The embodiments of the present invention will be described in further detail with reference to the drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
Fig. 1 is a schematic flow chart of a distance information completion method according to an embodiment of the present invention. Referring to fig. 1, an embodiment of the present invention provides a distance information completion method, including:
s1, determining a reference area based on the area to be compensated of the depth image; wherein the reference region is within the depth image.
It can be understood that, due to the fact that hardware of the 3D camera has defects or interference of a strong reflective object on an internal photosensitive element, a Depth image (Depth map, a distance imaging map representing an object to the camera) acquired by the 3D camera has a condition of missing part of distance points, so that real space position information of some objects in the image cannot be acquired, and subsequent tasks such as robot grabbing, navigation and ranging are affected. In the prior art, for completing the distance missing points, the most similar distance points around the missing points are directly adopted for supplementing, different distance points measured by ignoring the distance of an object from a camera in a 3D space have certain visual difference, and the supplemented distance points do not accord with the current effective distance. In addition, if the 3D data completion network model is used for completion, a large amount of data training needs to be performed on the GPU, and the method is a global point completion method, and cannot complete a specific point, consume excessive computing resources, and make model prediction time-consuming and unstable.
In the embodiment of the present invention, the region to be compensated is a region in which distance information compensation is required, and may be a region in which frame selection is performed for a certain object, a certain component, or a certain object part. For the region to be compensated selected by the frame, the embodiment of the invention performs comprehensive calculation by using the distance point information of the reference region inside or outside the region to be compensated, and performs distance information compensation on the distance missing point in the region to be compensated. The region to be compensated may be a circle, a polygon, an irregular figure, or the like.
In the embodiment of the present invention, the reference area may be one or more. For the determination of the reference region, the central point of the region to be compensated may be used as the central point of the reference region to be acquired, and the region reduced by several times of the region to be compensated may be used as the reference region based on the central point; or taking a plurality of small areas around the central point of the area to be compensated as a plurality of contraction areas. In addition, a region enclosed by two contours formed by amplifying the region to be compensated by different times can be respectively used as a reference region; or a plurality of small areas which are uniformly distributed around the outside of the area to be compensated can be taken as a plurality of reference areas. The specific reference area acquisition mode can be set according to actual requirements.
And S2, determining a candidate distance value according to the distance value of the effective distance point in the reference area. Wherein the effective distance point is a point where no distance value loss occurs.
In the embodiment of the present invention, for the determined reference region, the candidate distance value is obtained by performing calculation (may be average calculation, or may be weighted average by determining a weight value according to the distance between each distance point and the center point of the reference region) according to the distance value of the effective distance point in the reference region.
S3 completes the distance value for the distance missing point in the region to be completed based on the candidate distance value.
After the candidate distance value corresponding to the reference region is acquired, the distance value of the distance missing point in the region to be compensated may be compensated based on the candidate distance value according to different situations. For example, when there are a plurality of reference regions, an average value of candidate distance values corresponding to the plurality of reference regions may be used as the distance missing point in the region to be compensated for distance value, or corresponding weights may be set for the candidate distance values of the plurality of reference regions for weighted average, and a value obtained by weighted average may be used as the distance missing point in the region to be compensated for distance value.
According to the distance information completion method provided by the embodiment of the invention, the information completion of the distance missing points is carried out on the specific area in the image, model training and global completion are not required to be carried out on the whole depth image, a large amount of computing resources and computing time are saved, and the distance information completion efficiency is improved; meanwhile, the candidate distance value is comprehensively calculated according to the plurality of effective distance points of the reference area corresponding to the area to be compensated, so that the problem of low accuracy caused by the fact that a single distance point near a target point is adopted to supplement the distance value in the prior art is solved, and the accuracy of distance information completion of the distance missing point in the area to be compensated is improved.
In one embodiment, the reference region comprises a constricted region and an expanded region; wherein the contraction region and the expansion region are both within the depth image, the contraction region is located inside the region to be compensated, and the expansion region is located outside the region to be compensated;
step S2 may specifically include:
s21, determining a first candidate distance value according to the distance value of the effective distance point in the contraction area, and determining a second candidate distance value according to the distance value of the effective distance point in the expansion area.
Step S3 may specifically include:
and S31, performing distance value completion on the distance missing points in the area to be completed based on the first candidate distance value and the second candidate distance value.
In the embodiment of the present invention, the first candidate distance value and the second candidate distance value are calculated by simultaneously acquiring the contracted area and the expanded area of the area to be compensated, respectively, and the candidate distance values are calculated by comprehensively considering the effective distance points of the inner and outer reference areas of the area to be compensated. The determination of the contraction area may be performed by using a central point of the area to be compensated as a central point of the contraction area, and using an area reduced by a plurality of times of the area to be compensated as the contraction area; or taking a plurality of small areas around the central point of the area to be compensated as a plurality of contraction areas. For the expansion area, an area enclosed by two contours formed by amplifying the area to be compensated by different times can be respectively adopted as the expansion area; or a plurality of small areas which are uniformly distributed around the outside of the area to be compensated can be taken as a plurality of expansion areas. The number of the contraction areas and the expansion areas can be one or more, and the specific acquisition mode of the contraction areas and the expansion areas can be defined according to actual requirements.
In the embodiment of the present invention, for the determined contracted area and expanded area, a first candidate distance value and a second candidate distance value are obtained by performing calculation (average calculation may be performed, or weighted average may be performed by determining a weight value according to the distance between each distance point and the center point of the area) according to the distance values of the effective distance points in the two areas; if the number of the contraction areas is multiple, further carrying out average calculation or weighted average according to the candidate distance values corresponding to the plurality of contraction areas so as to calculate to obtain a first candidate distance value; similarly, if there are a plurality of expansion areas, the first candidate distance value is calculated by further performing an average calculation or a weighted average based on the candidate distance values corresponding to the plurality of expansion areas.
After the first candidate distance value corresponding to the contracted area and the second candidate distance value corresponding to the expanded area are acquired, the distance value of the distance missing point in the area to be compensated may be compensated based on the first candidate distance value and the second candidate distance value, depending on the case. For example, an average value of the first candidate distance value and the second candidate distance value may be used as the distance missing point in the area to be compensated, or a weighted average may be performed by setting a corresponding weight to the first candidate distance value and the second candidate distance value, and a value obtained by the weighted average may be used as the distance missing point in the area to be compensated.
According to the distance information completion method provided by the embodiment of the invention, the candidate distance value is comprehensively calculated according to the plurality of effective distance points of the contraction area and the expansion area corresponding to the area to be completed, and the distance information completion is carried out by comprehensively considering the inner reference area and the outer reference area of the area to be completed, so that the accuracy of distance information completion on the distance missing points in the area to be completed is further improved.
In one embodiment, step S31 may include the steps of:
s311, determining that the difference value between the first candidate distance value and the second candidate distance value is larger than a preset difference threshold value, and performing distance value completion on the distance missing point in the area to be completed by adopting the first candidate distance value;
s312, determining that a difference between the first candidate distance value and the second candidate distance value is not greater than the difference threshold, and performing distance value completion on the distance missing point in the to-be-completed region by using an average value of the first candidate distance value and the second candidate distance value.
In the embodiment of the present invention, the distance value of the distance missing point in the region to be compensated may be compensated in different cases based on the difference between the first candidate distance value and the second candidate distance value.
Specifically, when the difference value between the first candidate distance value and the second candidate distance value is greater than the preset difference threshold value, it is considered that the reference value of the distance point information of the expansion area outside the area to be supplemented for distance completion of the area to be supplemented is small, and therefore the distance value completion is performed on the distance missing point of the area to be supplemented by directly adopting the first candidate distance value corresponding to the contraction area.
Correspondingly, when the difference between the first candidate distance value and the second candidate distance value is not greater than the difference threshold, it is determined that the distance point information of the expanded region outside the region to be supplemented has a certain reference value for distance supplementation of the region to be supplemented, and therefore, it is necessary to perform an average operation on the first candidate distance value and the second candidate distance value to supplement the distance value of the distance missing point of the region to be supplemented.
According to the distance information completion method provided by the embodiment of the invention, the distance value completion is carried out on the distance missing points in the region to be completed according to the difference value between the first candidate distance value and the second candidate distance value and different conditions, so that the distance information to be completed in the region to be completed is closer to the actual scene, and the accuracy of the distance information completion is further improved.
In another embodiment, a reference value weight of the expanded region located outside the region to be compensated with respect to the distance to be compensated may be determined based on a difference between the first candidate distance value and the second candidate distance value, and the calculation weight of the first candidate distance value and the second candidate distance value may be determined based on the reference value weight. For example, different threshold intervals are preset, each threshold interval corresponds to a different internal and external reference area weight ratio, and a target threshold interval corresponding to the difference value between the first candidate distance value and the second candidate distance value is determined, so that the first candidate distance value and the second candidate distance value can be subjected to weighted average calculation by matching the corresponding internal and external reference area weight ratios to complete the distance information of the distance missing point.
In one embodiment, the region to be complemented is a polygon;
step S1 may include the steps of:
s11, determining an extension point on at least one edge line of the region to be compensated, determining a plurality of vertexes of the contraction region based on each extension point and a preset adduction distance, and obtaining the contraction region according to the vertexes;
and S12, determining the expansion area center point corresponding to each extension point based on each extension point and a preset outward expansion distance, and obtaining the expansion area according to the expansion area center point.
It should be noted that, for the frame selection of the region to be compensated, a polygon may be used for marking, and further, a rectangle may be used for marking. For the contraction area, the central points of four side lines of the rectangle may be used as extension points, and the extension points extend from the extension points (preferably in the direction perpendicular to the side lines) to the inside of the contraction area by a preset adduction distance to obtain four extending end points, and the four end points are connected to obtain a diamond-shaped area, which is the contraction area. For the expansion area, similarly, according to the extension points on the four edge lines of the rectangle, extending from the extension points (preferably in a direction perpendicular to the edge lines) to the outside of the area to be compensated by a preset flaring distance to obtain four extending end points, and using the end points as central points, and then according to a preset side length, forming and obtaining the corresponding four expansion areas. It should be noted that the adduction distance (the outward expansion distance) corresponding to different extending points may be different, and when the adduction distance corresponding to different extending points is different, the formed contraction area may be an irregular pattern. In other embodiments, one or more contraction zones and one or more expansion zones may be determined based on different numbers of extension points and different adduction or abduction distances.
According to the distance information completion method provided by the embodiment of the invention, the polygon is adopted to mark the region to be completed, the contraction region and the expansion regions are determined based on the side line of the polygonal region to be completed and the preset internal contraction distance or external expansion distance, so that a plurality of relevant regions near the region to be completed can be uniformly obtained, the first candidate distance value and the second candidate distance value are comprehensively calculated according to the distance effective values of the relevant regions, and the accuracy of distance information completion is further improved.
In one embodiment, after step S1, before step S21, the method further comprises the steps of:
s13, determining that the number of effective distance points in the contraction area is less than a preset first threshold value, and re-determining the contraction area by adjusting the adduction distance;
and S14, determining that the number of the effective distance points in the expansion area is less than a preset second threshold value, and re-determining the expansion area by adjusting the flaring distance.
It can be understood that when the determined contraction area or expansion area range in step S1 is smaller, the number of valid distance points in the contraction area or expansion area is more likely to be smaller, even 0, so to ensure that enough valid distance points can be obtained for candidate distance value calculation, the embodiment of the present invention determines the contraction area again by setting the first threshold and the second threshold, and when the number of valid distance points in the contraction area is less than the first threshold, adjusting the adduction distance (reducing the adduction distance can obtain a larger contraction area); when the number of the effective distance points in at least one expansion area is less than a preset second threshold, the expansion areas are determined again by adjusting the outward expansion distance (the expansion area farther away from the area to be compensated can be obtained by expanding the outward expansion distance), and the range of the expansion area can also be expanded by adjusting the side length.
According to the distance information complementing method of the embodiment of the invention, when the effective distance points in the contraction area or the expansion area are determined to be less, the distance information of the relevant area is considered to be insufficient for calculating the first candidate distance value or the second candidate distance value, so that the contraction area or the expansion area is re-determined by adjusting the degree of adduction or abduction, the first candidate distance value and the second candidate distance value of the area to be complemented are calculated by acquiring enough distance information, and the accuracy of distance information complementing is further improved.
In one embodiment, step S21 may include the steps of:
s21, determining a first distance average value of the contraction area, and determining the first candidate distance value according to the first distance average value;
s22, determining a second distance average value of the expansion area, and determining the second candidate distance value according to the second distance average value;
the first distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the contraction area; and the second distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the expansion area.
In the embodiment of the present invention, the first candidate distance value and the second candidate distance value are obtained by averaging, and specifically, in the case where there is only one shrinkage area and one expansion area, the average value is calculated as the first candidate distance value or the second candidate distance value by using the distance values of all the valid distance points in the shrinkage area or in the expansion area; in the case where there are a plurality of contraction areas or expansion areas, the distance average of all the effective distance points in each contraction area (expansion area) is calculated, and then the second candidate distance value is obtained by performing calculation (which may be averaging, or weighted averaging by determining different weights from areas in different directions) based on the distance average of the plurality of contraction areas (expansion areas).
According to the distance information completion method provided by the embodiment of the invention, the average value of the effective distance values of the relevant areas is used as the basis for calculating the candidate distance values, wherein for a plurality of contraction or expansion areas, the average value of the effective distance values of each area is firstly obtained respectively, and then the candidate distance values are calculated according to the average values of the distances of the areas, so that the distance point information of the relevant areas around the area to be completed can be uniformly obtained, and the universality and the accuracy of distance information completion are further improved.
In one embodiment, before step S1, the method may further include the steps of:
s15, determining the region to be compensated from the depth image based on the two-dimensional coordinate information of the target identification region of the color imaging image;
wherein the coordinate system of the color imaging graph and the coordinate system of the depth image are registered with each other in advance; the target identification area is an area where an object to be detected is located.
It should be noted that, in practical applications, distance information of a target scene is generally obtained by combining a color imaging map and a depth image, and since the color imaging map can more quickly and accurately identify a region of a target object than the depth image, coordinate systems of the color imaging map and the depth image may be mutually registered, then, the color imaging map is used to perform target identification to obtain two-dimensional coordinate information (e.g., four vertex coordinates of a rectangular frame) of a target identification region, and then, a corresponding region to be compensated may be framed in the depth image according to the same coordinates.
According to the distance information complementing method provided by the embodiment of the invention, the region to be complemented does not need to be directly obtained through the depth image, but the region to be complemented in the depth image is determined through the target identification region of the adaptive color imaging image, and the region of the target object can be identified more quickly and accurately compared with the depth image through the color imaging image, so that the efficiency and the accuracy of distance information complementing are further improved on the whole.
In one embodiment, before step S15, the method may further include the steps of:
s16, inputting the color imaging graph into a target recognition model, and obtaining the color imaging graph which is output by the target recognition model and marked with a target recognition area;
the target recognition model is obtained by training based on a sample color imaging graph and a target recognition area result corresponding to the sample color imaging graph.
It should be noted that, for the target recognition in the color imaging graph, a pre-trained target recognition model (such as fast-rcnn series, yolo series, etc.) can be used for recognition. For the training of the target recognition model, a sample color imaging graph with target objects (such as cups, boxes and the like) of different forms can be adopted, and then the model training is carried out by combining a marked target object frame to obtain the target recognition model.
According to the distance information completion method provided by the embodiment of the invention, the trained target recognition model is adopted to recognize the target object region in the color imaging image, so that the accuracy of selecting the region to be completed by the frame can be further improved, and the accuracy of distance information completion is further improved.
Referring to fig. 2 to 3, based on the above schemes, in order to better understand the distance information completing method provided by the embodiment of the present invention, the following specific examples are listed for detailed description:
in order to solve the problems of low efficiency and low accuracy in the prior art of completing distance missing points, the distance information completing method provided by the embodiment of the invention estimates the target distance points by adopting the partial distance points of the contraction area of the area to be completed, so that the single distance is not relied on, certain redundancy is achieved, meanwhile, the partial distance points of the expansion area of the area to be completed are adopted to estimate the target distance points, and the estimated distance points in the area can be corrected by utilizing the structural characteristics in space, so that more accurate target distance points are obtained for completing. It should be noted that, the completion distance points obtained by performing multi-directional calculation on the spatial structure are adopted, the powerful GPU is not relied on, additional calculation resources do not need to be consumed, the estimated target distance points are more stable, and the migration to various recognition algorithm frames is facilitated. The embodiment of the invention comprises the following specific steps:
1) the current scene data is captured by a 3D camera or the like, and a registered RGB map (color imaging map) and Depth map (Depth image) are obtained, which are denoted by S1 and S2, respectively.
2) The currently acquired RGB images are trained and predicted using a target recognition model, which may be any standard 2D target detection network (e.g., the faster-rcnn series, yolo series, etc.),a rectangular box (or other figures, a rectangle is used as an example in this example) can be predicted around the target, and the area where the target object is located is marked by using the rectangular box. The embodiment of the invention does not limit the type of the 2D target detection network. The upper left corner of the RGB image is taken as the origin of coordinates, the RGB image is taken as the positive direction of an x axis from left to right and the positive direction of a y axis from top to bottom, and the coordinates of the corner points of the rectangular frame containing the target object region are obtained through prediction
Wherein
And
respectively represent the coordinates of the upper left corner of the rectangular box,
and
respectively, the coordinates of the lower right corner of the rectangular frame. Similarly, a coordinate system which is the same as that in the RGB image is set on the Depth image, so that the corner point B of the rectangular frame can be directly migrated into the Depth image, and the region to be supplemented in the Depth image is obtained. Each coordinate value in the Depth map corresponds to an actual distance point P, so that a target distance point P can be selected in a to-be-compensated area of the Depth map * 。
3) And calculating a candidate point for distance information completion of the missing point in the region to be completed based on the contraction region and the expansion region corresponding to the region to be completed according to the Depth map and the coordinates of the region to be completed. The specific region solution is shown in figure 3 of the accompanying drawings.
a) The candidate distance value of the shrinkage area range is calculated. Firstly, the middle points of four sides of a rectangular frame of a region to be compensated are calculated
And
four contraction factors (adduction distances) [ cg ] are set 0 >0,cg 1 >0,cg 2 >0,cg 3 >0]A new contracted diamond-shaped area can be obtained, and four corner points of the diamond-shaped area are respectively
And
wherein
And
the candidate distance values calculated by adopting the shrunk diamond-shaped area can effectively filter out irrelevant areas compared with a method of directly shrinking a rectangle. Set the distance value set of the contracted diamond region points as
Calculating a candidate distance value corresponding to the shrinkage region by using the following formula
Wherein the content of the first and second substances,
k is
The number of all valid distance points (distance values not missing) present in the set. When k is 0 (which may be set to be less than a preset value), that is, there is no effective distance point in the region, the puncturing factor [ cg ] may be adjusted 0 ,cg 1 ,cg 2 ,cg 3 ]To obtain a new diamond region for calculation.
By adopting the same calculation mode, the vertex of the contracted diamond-shaped area can be connected with the extension point on the sideline of the area to be compensated to form an effective point set of the areas with four directions between the diamond-shaped area and the area to be compensated
And
and calculating candidate distances of each set, and averaging to obtain a final candidate distance value.
b) Candidate distance values of the expanded region range are calculated. The middle points of the four sides of the rectangular frame in the step a) are
And
setting four expansion factors (outward expansion distances) [ ep ] according to the current scene 0 >0,ep 1 >0,ep 2 >0,ep 3 >0]Obtaining coordinates of the expanded distance points
And
wherein
And
the invalid missing value or mutation value with the distance value of 0 caused by objective factors such as equipment, angle and the like of the distance point selected after expansion is prevented from appearing is respectively set
As a center, the distance values of four expansion region points with side length d are collected as
And
calculating a candidate distance value corresponding to the expanded region by using the following formula
Wherein the content of the first and second substances,
m is
The number of all effective distance points in the set;
n is
The number of all effective distance points in the set;
w is
The number of all effective distance points in the set;
u is
The number of all valid distance points in the set. When m, n, w, u are 0 (can be set as preset values), the expansion factor [ ep ] can be adjusted 0 ,ep 1 ,ep 2 ,ep 3 ](or adjust the side length) to obtain a new expansion region for calculation. Candidate distance value calculated by four-direction synthesis
The spatial information around the target can be effectively utilized to obtain more accurate distance.
4) And completing the distance value of the distance missing points in the region to be completed. Firstly, judging a target distance point P directly acquired by equipment * If a distance value of (b) is present, when P * If the distance value (distance missing point) does not exist, the candidate distance value is used for calculation. Setting a threshold value phi>0, when
The target distance point
When in use
Target distance point
Target distance point P missing all distance values in region to be complemented * Completing the completion.
It should be noted that, the above embodiment describes a scheme of implementing distance information completion by simultaneously referring to effective distance values of the contraction area and the expansion area, and in other embodiments, distance information completion may also be implemented by separately referring to effective distance values of the contraction area, where the acquisition of the contraction area may refer to the contraction area determination manner described in the above embodiment; in addition, the distance information completion may also be implemented by referring to the effective distance value of the expansion region alone, wherein the acquisition of the expansion region may refer to the shrinkage region determination manner described in the above embodiment.
It should be noted that the embodiment of the present invention is applicable to a plurality of deployments of platforms with visual perception: smart phones, robots (such as home service robots), surveillance cameras, smart appliances, and the like. Data collected by various sensors may be used as input: the system comprises an RGB-D infrared camera, an RGB camera + laser radar, an RGB camera + millimeter wave radar, an RGB camera + infrared camera and other elements capable of performing RGB imaging and ranging imaging.
Compared with the prior art, the embodiment of the invention has the following beneficial effects:
1) the embodiment of the invention has stable and fast calculation, does not depend on high-performance calculation resources, and is convenient to be transplanted into equipment.
2) The embodiment of the invention can realize plug and play and can be flexibly adapted to the existing recognition algorithm.
3) The embodiment of the invention utilizes the distance points of the specific area to complete, can freely adjust the specific area aiming at different scene tasks, and has certain universality.
4) The embodiment of the invention does not depend on the imaging performance of the 3D camera too much, and the cost of hardware is saved.
Referring to fig. 4, fig. 4 is a schematic block diagram of a distance information completing device according to an embodiment of the present invention, where the distance information completing device according to the embodiment of the present invention includes:
the region determining module 1 is configured to determine a reference region based on a region to be compensated of the depth image;
the information candidate module 2 is used for determining a candidate distance value according to the distance value of the effective distance point in the reference area;
an information completion module 3, configured to complete distance values of the distance missing points in the to-be-completed region based on the candidate distance values;
wherein the reference region is within the depth image; the effective distance point is a point where no distance value loss occurs.
In one embodiment, the reference region comprises a constricted region and an expanded region; wherein the contraction region and the expansion region are both within the depth image, the contraction region is located inside the region to be compensated, and the expansion region is located outside the region to be compensated;
the information candidate module 2 is specifically configured to:
determining a first candidate distance value according to the distance value of the effective distance point in the contraction area, and determining a second candidate distance value according to the distance value of the effective distance point in the expansion area;
the information complementing module 3 is specifically configured to:
and performing distance value completion on the distance missing points in the region to be completed based on the first candidate distance value and the second candidate distance value.
In one embodiment, the information completing module 3 is specifically configured to:
determining that a difference value between the first candidate distance value and the second candidate distance value is greater than a preset difference threshold value, and performing distance value completion on the distance missing point in the area to be completed by adopting the first candidate distance value;
and determining that the difference value between the first candidate distance value and the second candidate distance value is not greater than the gap threshold value, and performing distance value completion on the distance missing point in the region to be completed by adopting the average value of the first candidate distance value and the second candidate distance value.
In one embodiment, the region to be complemented is a polygon;
the region determining module 1 is specifically configured to:
determining an extension point on at least one edge line of the region to be compensated, determining a plurality of vertexes of the contraction region based on each extension point and a preset adduction distance, and obtaining the contraction region according to the vertexes;
and determining an expansion area central point corresponding to each extension point based on each extension point and a preset outward expansion distance, and obtaining the expansion area according to the expansion area central point.
In one embodiment, the apparatus further comprises a region adjustment module to:
determining that the number of effective distance points in the contraction area is less than a preset first threshold value, and re-determining the contraction area by adjusting the adduction distance;
and determining that the number of effective distance points in the expansion area is less than a preset second threshold value, and re-determining a plurality of expansion areas by adjusting the outward expansion distance.
In one embodiment, the information candidate module 2 is specifically configured to:
determining a first distance average of the shrinkage region, and determining the first candidate distance value according to the first distance average;
determining a second distance average for the expanded region and determining the second candidate distance value based on the second distance average;
the first distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the contraction area; and the second distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the expansion area.
In one embodiment, the apparatus further comprises a region matching module to:
determining the region to be compensated from the depth image based on two-dimensional coordinate information of a target identification region of a color imaging image;
wherein the coordinate system of the color imaging graph and the coordinate system of the depth image are registered with each other in advance; the target identification area is an area where an object to be detected is located.
In one embodiment, the apparatus further comprises a target identification module to:
inputting the color imaging graph into a target recognition model to obtain a color imaging graph which is output by the target recognition model and marked with a target recognition area;
the target recognition model is obtained by training based on a sample color imaging graph and a target recognition area result corresponding to the sample color imaging graph.
Fig. 5 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 5: a processor (processor)510, a communication Interface (Communications Interface)520, a memory (memory)530 and a communication bus 540, wherein the processor 510, the communication Interface 520 and the memory 530 communicate with each other via the communication bus 540. Processor 510 may call logic instructions in memory 530 to perform the following method:
s1, determining a reference area based on the area to be compensated of the depth image;
s2, determining a candidate distance value according to the distance value of the effective distance point in the reference area;
s3, completing the distance value of the distance missing point in the area to be completed based on the candidate distance value;
wherein the reference region is within the depth image; the effective distance point is a point where no distance value loss occurs.
Furthermore, the logic instructions in the memory 530 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. 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.
In another aspect, an embodiment of the present invention discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, when the program instructions are executed by a computer, the computer can perform the methods provided by the above-mentioned method embodiments, for example, including:
s1, determining a reference area based on the area to be compensated of the depth image;
s2, determining a candidate distance value according to the distance value of the effective distance point in the reference area;
s3, completing the distance value of the distance missing point in the area to be completed based on the candidate distance value;
wherein the reference region is within the depth image; the effective distance point is a point where no distance value loss occurs.
In another aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented to perform the transmission method provided in the foregoing embodiments when executed by a processor, for example, the method includes:
s1, determining a reference area based on the area to be compensated of the depth image;
s2, determining a candidate distance value according to the distance value of the effective distance point in the reference area;
s3, completing the distance value of the distance missing point in the area to be completed based on the candidate distance value;
wherein the reference region is within the depth image; the effective distance point is a point where no distance value loss occurs.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on such understanding, the above technical solutions substantially or contributing to the related art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments.
Finally, it should be noted that the above embodiments are only for illustrating the present invention, and do not limit the present invention. Although the present invention has been described in detail with reference to the embodiments, it should be understood by those skilled in the art that various combinations, modifications or equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention, and the technical solution of the present invention is covered by the claims of the present invention.
Claims (11)
1. A method for complementing distance information, comprising:
determining a reference region based on a region to be compensated of the depth image;
determining a candidate distance value according to the distance value of the effective distance point in the reference area;
completing the distance value of the distance missing point in the area to be completed based on the candidate distance value;
wherein the reference region is within the depth image; the effective distance point is a point where the distance value loss does not occur.
2. The distance information complementing method of claim 1, wherein the reference area includes a contraction area and an expansion area; wherein the contraction region and the expansion region are both within the depth image, the contraction region is located inside the region to be compensated, and the expansion region is located outside the region to be compensated;
the determining a candidate distance value according to the distance value of the effective distance point in the reference region includes:
determining a first candidate distance value according to the distance value of the effective distance point in the contraction area, and determining a second candidate distance value according to the distance value of the effective distance point in the expansion area;
the distance value completion of the distance missing points in the region to be completed based on the candidate distance values includes:
and performing distance value completion on the distance missing points in the region to be completed based on the first candidate distance value and the second candidate distance value.
3. The distance information complementing method of claim 2, wherein the distance value complementing the distance missing point in the area to be complemented based on the first candidate distance value and the second candidate distance value comprises:
determining that a difference value between the first candidate distance value and the second candidate distance value is greater than a preset difference threshold value, and performing distance value completion on the distance missing point in the area to be completed by adopting the first candidate distance value;
and determining that the difference value between the first candidate distance value and the second candidate distance value is not greater than the gap threshold value, and performing distance value completion on the distance missing point in the region to be completed by adopting the average value of the first candidate distance value and the second candidate distance value.
4. The distance information complementing method of claim 2, wherein the area to be complemented is a polygon;
the determining of the reference region based on the region to be compensated of the depth image includes:
determining an extension point on at least one edge line of the region to be compensated, determining a plurality of vertexes of the contraction region based on each extension point and a preset adduction distance, and obtaining the contraction region according to the vertexes;
and determining an expansion area central point corresponding to each extension point based on each extension point and a preset outward expansion distance, and obtaining the expansion area according to the expansion area central point.
5. The distance information complementing method of claim 4, wherein after determining the reference region based on the depth image to-be-complemented region, before determining a first candidate distance value based on the distance values of the valid distance points within the contracted region and determining a second candidate distance value based on the distance values of the valid distance points within the expanded region, further comprising:
determining that the number of effective distance points in the contraction area is less than a preset first threshold value, and re-determining the contraction area by adjusting the adduction distance;
and determining that the number of the effective distance points in the expansion area is less than a preset second threshold value, and re-determining the expansion area by adjusting the flaring distance.
6. The distance information complementing method of claim 4, wherein determining a first candidate distance value based on the distance values of the valid distance points within the contracted area and determining a second candidate distance value based on the distance values of the valid distance points within the expanded area comprises:
determining a first distance average of the shrinkage region, and determining the first candidate distance value according to the first distance average;
determining a second distance average for the expanded region and determining the second candidate distance value based on the second distance average;
the first distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the contraction area; and the second distance average value is obtained by carrying out average operation according to the distance values of all effective distance points in the expansion area.
7. The distance information complementing method of claim 1, further comprising, before determining the reference region based on the depth image to-be-complemented region:
determining the region to be compensated from the depth image based on two-dimensional coordinate information of a target identification region of a color imaging image;
wherein the coordinate system of the color imaging graph and the coordinate system of the depth image are registered with each other in advance; the target identification area is an area where an object to be detected is located.
8. The distance information complementing method of claim 7, further comprising, before determining the region to be complemented from the depth image based on two-dimensional coordinate information of the target identification region of the color imaging map,:
inputting the color imaging graph into a target recognition model to obtain a color imaging graph which is output by the target recognition model and marked with a target recognition area;
the target recognition model is obtained by training based on a sample color imaging graph and a target recognition area result corresponding to the sample color imaging graph.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the distance information complementing method of any one of claims 1 to 8 when executing the program.
10. A non-transitory computer-readable storage medium, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the distance information complementing method of any one of claims 1 to 8.
11. A computer program product comprising a computer program, wherein the computer program, when executed by a processor, implements the distance information completion method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210471068.2A CN114842057A (en) | 2022-04-28 | 2022-04-28 | Distance information complementing method, apparatus, storage medium, and computer program product |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210471068.2A CN114842057A (en) | 2022-04-28 | 2022-04-28 | Distance information complementing method, apparatus, storage medium, and computer program product |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114842057A true CN114842057A (en) | 2022-08-02 |
Family
ID=82567245
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210471068.2A Pending CN114842057A (en) | 2022-04-28 | 2022-04-28 | Distance information complementing method, apparatus, storage medium, and computer program product |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114842057A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116524201A (en) * | 2023-03-29 | 2023-08-01 | 锋睿领创(珠海)科技有限公司 | Feature extraction method, device, equipment and medium of multi-scale gating fusion unit |
-
2022
- 2022-04-28 CN CN202210471068.2A patent/CN114842057A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116524201A (en) * | 2023-03-29 | 2023-08-01 | 锋睿领创(珠海)科技有限公司 | Feature extraction method, device, equipment and medium of multi-scale gating fusion unit |
| CN116524201B (en) * | 2023-03-29 | 2023-11-17 | 锋睿领创(珠海)科技有限公司 | Feature extraction method, device, equipment and medium of multi-scale gating fusion unit |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110826357B (en) | Method, device, medium and equipment for three-dimensional detection and intelligent driving control of object | |
| US8903161B2 (en) | Apparatus for estimating robot position and method thereof | |
| US20210042929A1 (en) | Three-dimensional object detection method and system based on weighted channel features of a point cloud | |
| CN108875683B (en) | Robot vision tracking method and system | |
| US8199977B2 (en) | System and method for extraction of features from a 3-D point cloud | |
| CN111582054B (en) | Point cloud data processing method and device and obstacle detection method and device | |
| CN110493488B (en) | Video image stabilization method, video image stabilization device and computer readable storage medium | |
| CN107844750A (en) | A kind of water surface panoramic picture target detection recognition methods | |
| CN109472820B (en) | Monocular RGB-D camera real-time face reconstruction method and device | |
| CN108322724B (en) | Image solid matching method and binocular vision equipment | |
| CN113052907B (en) | Positioning method of mobile robot in dynamic environment | |
| CN111998862B (en) | BNN-based dense binocular SLAM method | |
| CN112883820A (en) | Road target 3D detection method and system based on laser radar point cloud | |
| CN111553946A (en) | Method and device for removing ground point cloud and obstacle detection method and device | |
| CN112967340A (en) | Simultaneous positioning and map construction method and device, electronic equipment and storage medium | |
| CN114842057A (en) | Distance information complementing method, apparatus, storage medium, and computer program product | |
| CN115533902A (en) | Visual guidance-based unstacking method and device, electronic equipment and system | |
| CN117011660A (en) | Dot line feature SLAM method for fusing depth information in low-texture scene | |
| CN114897999B (en) | Object pose recognition method, electronic device, storage medium, and program product | |
| CN114972491A (en) | Visual SLAM method, electronic device, storage medium and product | |
| US20230419605A1 (en) | Map generation apparatus, map generation method, and non-transitory computer-readable medium storing program | |
| CN114399800A (en) | Human face posture estimation method and device | |
| CN114693546A (en) | Image denoising method and device, electronic equipment and computer readable storage medium | |
| CN113887289A (en) | Monocular three-dimensional object detection method, device, equipment and product | |
| CN112862865A (en) | Detection and identification method and device for underwater robot and computer 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 |