CN108399632B - An RGB-D camera depth image inpainting method for joint color images - Google Patents

An RGB-D camera depth image inpainting method for joint color images Download PDF

Info

Publication number
CN108399632B
CN108399632B CN201810174885.5A CN201810174885A CN108399632B CN 108399632 B CN108399632 B CN 108399632B CN 201810174885 A CN201810174885 A CN 201810174885A CN 108399632 B CN108399632 B CN 108399632B
Authority
CN
China
Prior art keywords
depth
rgb
camera
cavity
color
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
Application number
CN201810174885.5A
Other languages
Chinese (zh)
Other versions
CN108399632A (en
Inventor
张毅
戴健春
张瀚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN201810174885.5A priority Critical patent/CN108399632B/en
Publication of CN108399632A publication Critical patent/CN108399632A/en
Application granted granted Critical
Publication of CN108399632B publication Critical patent/CN108399632B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

本发明涉及一种联合彩色图像的RGB‑D相机深度图像修复方法，属于深度图像修复领域。该方法包括：采用张氏标定法对彩色、深度摄像头进行标定，获取相机的内外参数；根据小孔成像原理以及坐标变换实现彩色、深度摄像头坐标对齐；分别获取场景彩色、深度图像，采用深度阈值法将其二值化；判断空洞连通域的大小以确定存在空洞；进行膨胀操作获取空洞邻域；计算空洞领域像素点的深度值方差，根据其大小将空洞分为遮挡、平面内空洞；分别采用颜色一致性和领域相似性对两类空洞进行修复；对修复后的图像采用局部滤波法进行滤波，去除噪声。本发明在保证深度图像修复的同时，尽可能的保持了深度图像的原始信息，并提高了深度图像的修复效率。

Figure 201810174885

The invention relates to an RGB-D camera depth image restoration method combined with color images, belonging to the field of depth image restoration. The method includes: calibrating the color and depth cameras by Zhang's calibration method, and obtaining the internal and external parameters of the cameras; realizing the coordinate alignment of the color and depth cameras according to the principle of pinhole imaging and coordinate transformation; respectively obtaining the scene color and depth images, and using a depth threshold Binarize it by the method; judge the size of the connected domain of the hole to determine the existence of a hole; perform the expansion operation to obtain the neighborhood of the hole; The two types of holes are repaired by color consistency and domain similarity; the repaired image is filtered by local filtering to remove noise. The invention keeps the original information of the depth image as much as possible while ensuring the restoration of the depth image, and improves the restoration efficiency of the depth image.

Figure 201810174885

Description

RGB-D camera depth image restoration method based on color image combination
Technical Field
The invention belongs to the field of depth image restoration, and relates to an RGB-D depth image restoration method combining color images.
Background
With the development of vision technology, more and more kinds of cameras are applied to display scenes, for example: the common color camera is used for face recognition, the CCD camera is used for obstacle detection, and the Kinect is used for human skeleton tracking and the like. The common two-dimensional plane camera can acquire scene two-dimensional information, and for a three-dimensional display space, the two-dimensional plane camera obviously cannot meet the requirements of practical application. This situation is primarily improved with the advent of RGB-D cameras, such as Kinect published by microsoft, which is very convenient for acquiring depth information while acquiring color images of a scene. With the development of microsoft kinect (natal), the depth camera attracts more and more people's eyes, and can be used in the fields of human body tracking, three-dimensional reconstruction, human-computer interaction, SLAM and the like. However, the depth image obtained by the RGB-D camera generally contains many holes, and the depth information cannot be directly applied to the real application, which greatly affects the application range of the RGB-D camera.
Therefore, an effective depth image restoration method is needed to solve the problem that depth image information obtained by RGB-D has holes and cannot be directly used.
Disclosure of Invention
In view of the above, an object of the present invention is to provide a depth image restoration method for an RGB-D camera combined with a color image, which provides an effective depth image restoration method for solving the problem that depth image information obtained by RGB-D has holes and cannot be directly used.
In order to achieve the purpose, the invention provides the following technical scheme:
a depth image restoration method of an RGB-D camera combining color images comprises the following steps:
s1: calibrating the color camera and the depth camera by adopting a classical Zhang calibration method to obtain internal and external parameters of the camera;
s2: realizing the coordinate alignment of the color camera and the depth camera according to the pinhole imaging principle and the coordinate transformation;
s3: respectively acquiring a scene color image and a depth image, and binarizing the depth image according to a depth threshold method;
s4: judging the size of a depth image hole connected domain to determine that a hole exists;
s5: performing expansion operation on the depth image to obtain a cavity neighborhood;
s6: calculating the variance of the depth values of the pixel points in the cavity field, and dividing the cavity into a shielding cavity and an in-plane cavity according to the variance;
s7: repairing the two types of cavities by adopting color consistency and field similarity according to different cavity types;
s8: and filtering the repaired image by adopting a local filtering method, and removing noise while ensuring the originality of the depth value.
Further, the step S2 specifically includes the following steps:
s21: the rotation matrix R of the color camera is obtained in step S1rgbAnd translation vector TrgbRotation matrix R of depth camerairAnd translation vector Tir(ii) a Let P be a point in the world coordinate system, PrgbAnd PirRespectively, obtaining a relation formula by a pinhole imaging model according to projection coordinate points of P under the color camera and the depth camera:
Prgb＝RrgbP+Trgb
Pir＝RirP+Tir
s22: let prgbAnd pirThe projection coordinates of the point on the RGB image plane and the depth image plane are respectively represented by an internal reference matrix A of the camerargbAnd AirObtaining:
prgb＝ArgbPrgb
pir＝AirPir
s23: the coordinates of the depth camera are different from the coordinates of the RGB camera, PrgbAnd PirA rotation matrix R and a translation matrix T are used for being related, and the relationship between the rotation matrix R and the translation matrix T is as follows:
Prgb＝RPir+T
s24: cancelling P by the S21 equation yields:
Figure BDA0001587003310000021
s25: the corresponding terms of simultaneous S23 and S24 equations are equal to obtain R and T:
Figure BDA0001587003310000022
Figure BDA0001587003310000023
s26: and (3) combining equations S22 and S23 to finally obtain the corresponding relation between the color image pixel point and the depth image pixel point as follows:
Figure BDA0001587003310000024
further, the step S3 specifically includes the following steps:
s31: simultaneously acquiring a scene image according to the aligned color camera and depth camera;
s32: the original depth data are quantized, the depth values are converted into gray levels of 0-255, and the quantization formula is as follows:
Figure BDA0001587003310000025
where depth is the depth value after quantization, DpFor each pixel point's depth value, DmaxIs the maximum depth value;
s33: and carrying out binarization on the obtained depth image according to a depth threshold value method, wherein a formula is expressed as follows:
Figure BDA0001587003310000031
wherein I (x, y) is the binarized image, depth (x, y) represents the depth value of the pixel point after quantization, DthrTo a binarized depth threshold.
Further, the step S4 specifically includes: after binarization processing of S3, judging the size of a cavity connected domain to distinguish space-time cavities and noise points; if the connected domain is judged to be a hole, continuing the subsequent processing; if not, the subsequent processing is not performed.
Further, the step S5 specifically includes the following steps:
s51: expanding operation H for the cavity judged at S4ep；
S52: obtaining void field HnpThe expression is as follows:
Hnp＝Hep-Ho
wherein HoIs a hollow region.
Further, the step S6 specifically includes the following steps:
s61: according to the hole neighborhood obtained in S5, calculating the variance delta of the depth values of the quantized pixel points in the neighborhooddepthThe calculation formula is as follows:
Figure BDA0001587003310000032
wherein DiThe depth value of the i pixel points in the hollow domain is represented,
Figure BDA0001587003310000033
representing the average value of depth values of the cavity field, i representing the number of pixel points, and N representing the total number of pixel points of the cavity field;
s62: neighborhood variance Δ calculated using S61depthAnd a void determination threshold value deltathBy comparison, if the depth variance Δ of the hole neighborhood isdepthLess than ΔthDetermining the hollow in the plane if the number is larger than or equal to deltathThen the occlusion hole is determined.
Further, the step S7 specifically includes: for the cavity judged to be in-plane, namely the cavity is generated by not reflecting infrared light in the object plane, because the cavity exists in a plane, the depth value of the cavity area is similar to the depth of the adjacent pixel points, and the repairing is carried out by utilizing the depth value alignment of the field pixel points; and for the holes judged to be shielded, the holes are generated due to mutual shielding of objects, and for the hole repairing, color images are used for repairing, namely, the depth values of pixel points in the hole neighborhood and the pixel points in the hole area, which have the color close to that of the color images, are found out for repairing.
Further, the step S8 specifically includes: because the hole repaired by the S7 still has fine noise, in order to ensure the originality of the depth data as much as possible and remove the noise, the local upper edge filtering method is adopted to perform image filtering.
The invention has the beneficial effects that: according to the method, the cavities are divided into two types according to a cavity neighborhood variance threshold method, the combined color image is used for repairing the cavities by respectively adopting domain similarity and color consistency, and the images after the cavities are repaired are subjected to noise removal by adopting a local filtering method. Compared with the traditional rough restoration method adopting the bilateral filtering algorithm, the method ensures the effectiveness of the restoration of the depth image and simultaneously ensures the originality of the depth image information as much as possible.
Drawings
In order to make the object, technical scheme and beneficial effect of the invention more clear, the invention provides the following drawings for explanation:
FIG. 1 is a system flow diagram of the present invention;
FIG. 2 is a schematic diagram of hole neighborhood calculation;
FIG. 3 is a schematic diagram of the formation of a blocking void.
Detailed Description
Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
FIG. 1 is a system flow chart of the present invention, as shown in FIG. 1, a method for repairing RGB-D camera depth image of combined color image, which comprises the steps of firstly realizing the alignment of coordinates of a color camera and a depth camera, carrying out binarization on the quantized depth image according to a depth threshold method, obtaining a cavity neighborhood through expansion operation, calculating a neighborhood variance to judge the type of the cavity, respectively repairing different cavities by adopting domain similarity and color consistency, and finally carrying out image filtering by adopting a local bilateral filtering method.
The method specifically comprises the following steps:
s1: calibrating the color camera and the depth camera by adopting a classical Zhang calibration method to obtain internal and external parameters of the camera;
s2: realizing the coordinate alignment of the color camera and the depth camera according to the pinhole imaging principle and the coordinate transformation;
s3: respectively acquiring a scene color image and a depth image, and binarizing the depth image according to a depth threshold method;
s4: judging the size of a depth image hole connected domain to determine that a hole exists: after binarization processing of S3, judging the size of a cavity connected domain to distinguish space-time cavities and noise points; if the connected domain is judged to be a hole, continuing the subsequent processing; if not, the subsequent processing is not performed.
S5: performing expansion operation on the depth image to obtain a cavity neighborhood;
s6: calculating the variance of the depth values of the pixel points in the cavity field, and dividing the cavity into a shielding cavity and an in-plane cavity according to the variance;
s7: repairing the two types of cavities by adopting color consistency and field similarity according to different cavity types;
for the cavity judged to be in-plane, namely the cavity is generated by not reflecting infrared light in the object plane, because the cavity exists in a plane, the depth value of the cavity area is similar to the depth of the adjacent pixel points, and the repairing is carried out by utilizing the depth value alignment of the field pixel points; for the holes judged to be blocked, the holes are generated due to mutual blocking of the objects, as shown in fig. 3, the holes are repaired by using the color images, that is, the depth values of the pixel points in the vicinity of the holes and in the holes, which are closer to the color of the color images, are found for repairing.
S8: filtering the repaired image by adopting a local filtering method, and removing noise while ensuring the originality of the depth value;
because the hole repaired by the S7 still has fine noise, in order to ensure the originality of the depth data as much as possible and remove the noise, the local upper edge filtering method is adopted to perform image filtering.
Step S2 specifically includes the following steps:
s21: the rotation matrix R of the color camera is obtained in step S1rgbAnd translation vector TrgbRotation matrix R of depth camerairAnd translation vector Tir(ii) a Let P be a point in the world coordinate system, PrgbAnd PirRespectively, obtaining a relation formula by a pinhole imaging model according to projection coordinate points of P under the color camera and the depth camera:
Prgb＝RrgbP+Trgb
Pir＝RirP+Tir
s22: let pgbrAnd pirThe projection coordinates of the point on the RGB image plane and the depth image plane are respectively represented by an internal reference matrix A of the camerargbAnd AirObtaining:
prgb＝ArgbPrgb
pir＝AirPir
s23: the coordinates of the depth camera are different from the coordinates of the RGB camera, PrgbAnd PirA rotation matrix R and a translation matrix T are used for being related, and the relationship between the rotation matrix R and the translation matrix T is as follows:
Prgb＝RPir+T
s24: cancelling P by the S21 equation yields:
Figure BDA0001587003310000051
s25: the corresponding terms of simultaneous S23 and S24 equations are equal to obtain R and T:
Figure BDA0001587003310000052
Figure BDA0001587003310000053
s26: combining the equations S22 and S23 to obtain the final colorThe corresponding relation between the image pixel point and the depth image pixel point is as follows:
Figure BDA0001587003310000054
step S3 specifically includes the following steps:
s31: simultaneously acquiring a scene image according to the aligned color camera and depth camera;
s32: the original depth data are quantized, the depth values are converted into gray levels of 0-255, and the quantization formula is as follows:
Figure BDA0001587003310000055
where depth is the depth value after quantization, DpFor each pixel point's depth value, DmaxIs the maximum depth value;
s33: and carrying out binarization on the obtained depth image according to a depth threshold value method, wherein a formula is expressed as follows:
Figure BDA0001587003310000061
wherein I (x, y) is the binarized image, depth (x, y) represents the depth value of the pixel point after quantization, DthrTo a binarized depth threshold.
Step S5 specifically includes the following steps:
s51: expanding operation H for the cavity judged at S4ep；
S52: obtaining void field HnpThe expression is as follows:
Hnp＝Hep-Ho
wherein HoIs a hollow region. As shown in fig. 2, a 3 × 3 cross-shaped structural element is used to perform an expansion operation on the hole, and the original hole image is subtracted from the expanded image to obtain a hole neighborhood.
Step S6 specifically includes the following steps:
s61: obtaining a hole neighborhood according to S5Calculating the variance delta of the depth value after the quantization of the pixel points in the neighborhooddepthThe calculation formula is as follows:
Figure BDA0001587003310000062
wherein DiThe depth value of the i pixel points in the hollow domain is represented,
Figure BDA0001587003310000063
representing the average value of depth values of the cavity field, i representing the number of pixel points, and N representing the total number of pixel points of the cavity field;
s62: neighborhood variance Δ calculated using S61depthAnd a void determination threshold value deltathBy comparison, if the depth variance Δ of the hole neighborhood isdepthLess than ΔthDetermining the hollow in the plane if the number is larger than or equal to deltathThen the occlusion hole is determined.
Finally, it is noted that the above-mentioned preferred embodiments illustrate rather than limit the invention, and that, although the invention has been described in detail with reference to the above-mentioned preferred embodiments, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the scope of the invention as defined by the appended claims.

Claims (5)

1. A depth image restoration method of an RGB-D camera combined with a color image is characterized by comprising the following steps:
s1: calibrating the color camera and the depth camera by adopting a classical Zhang calibration method to obtain internal and external parameters of the camera;
s2: realizing the coordinate alignment of the color camera and the depth camera according to the pinhole imaging principle and the coordinate transformation; the method specifically comprises the following steps:
s21: the rotation matrix R of the color camera is obtained in step S1rgbAnd translation vector TrgbRotation matrix R of depth camerairAnd translation vector Tir(ii) a Let P be a point in the world coordinate system, PrgbAnd PirRespectively, obtaining a relation formula by a pinhole imaging model according to projection coordinate points of P under the color camera and the depth camera:
Prgb＝RrgbP+Trgb
Pir＝RirP+Tir
s22: let prgbAnd pirThe projection coordinates of the point on the RGB image plane and the depth image plane are respectively represented by an internal reference matrix A of the camerargbAnd AirObtaining:
prgb＝ArgbPrgb
pir＝AirPir
s23: the coordinates of the depth camera are different from the coordinates of the RGB camera, PrgbAnd PirA rotation matrix R and a translation matrix T are used for being related, and the relationship between the rotation matrix R and the translation matrix T is as follows:
Prgb＝RPir+T
s24: cancelling P by the S21 equation yields:
Figure FDA0003043883360000011
s25: the corresponding terms of simultaneous S23 and S24 equations are equal to obtain R and T:
Figure FDA0003043883360000012
Figure FDA0003043883360000013
s26: and (3) combining equations S22 and S23 to finally obtain the corresponding relation between the color image pixel point and the depth image pixel point as follows:
Figure FDA0003043883360000014
s3: respectively acquiring a scene color image and a depth image, and binarizing the depth image according to a depth threshold method;
s4: judging the size of a depth image hole connected domain to determine that a hole exists;
s5: performing expansion operation on the depth image to obtain a cavity neighborhood; s6: calculating the variance of the depth values of the pixel points in the cavity field, and dividing the cavity into a shielding cavity and an in-plane cavity according to the variance; the method specifically comprises the following steps:
s61: according to the hole neighborhood obtained in S5, calculating the variance delta of the depth values of the quantized pixel points in the neighborhooddepthThe calculation formula is as follows:
Figure FDA0003043883360000021
wherein DiThe depth value of the i pixel points in the hollow domain is represented,
Figure FDA0003043883360000022
representing the average value of depth values of the cavity field, i representing the number of pixel points, N representing the total number of pixel points of the cavity field, and HnpIs a cavity field;
s62: neighborhood variance Δ calculated using S61depthAnd a void determination threshold value deltathBy comparison, if the depth variance Δ of the hole neighborhood isdepthLess than ΔthDetermining the hollow in the plane if the number is larger than or equal to deltathJudging the shielding cavity;
s7: repairing the two types of cavities by adopting color consistency and field similarity according to different cavity types; the method specifically comprises the following steps: for the cavity judged to be in-plane, namely the cavity is generated by not reflecting infrared light in the object plane, because the cavity exists in a plane, the depth value of the cavity area is similar to the depth of the adjacent pixel points, and the repairing is carried out by utilizing the depth value alignment of the field pixel points; for the holes judged to be shielded, the holes are generated due to mutual shielding of objects, and for the hole repairing, color images are used for repairing, namely, the depth values of pixel points in the hole neighborhood and the pixel points in the hole area, the color of which is close to that of the color images, are found out for repairing;
s8: and filtering the repaired image by adopting a local filtering method, and removing noise while ensuring the originality of the depth value.
2. The method for RGB-D camera depth image restoration combined with color image according to claim 1, wherein the step S3 specifically includes the following steps:
s31: simultaneously acquiring a scene image according to the aligned color camera and depth camera;
s32: the original depth data are quantized, the depth values are converted into gray levels of 0-255, and the quantization formula is as follows:
Figure FDA0003043883360000023
where depth is the depth value after quantization, DpFor each pixel point's depth value, DmaxIs the maximum depth value;
s33: and carrying out binarization on the obtained depth image according to a depth threshold value method, wherein a formula is expressed as follows:
Figure FDA0003043883360000024
wherein I (x, y) is the binarized image, depth (x, y) represents the depth value of the pixel point after quantization, DthrTo a binarized depth threshold.
3. The method for RGB-D camera depth image restoration combined with color image according to claim 1, wherein the step S4 specifically includes: after binarization processing of S3, judging the size of a cavity connected domain to distinguish space-time cavities and noise points; if the connected domain is judged to be a hole, continuing the subsequent processing; if not, the subsequent processing is not performed.
4. The method for RGB-D camera depth image restoration combined with color image according to claim 1, wherein the step S5 specifically includes the following steps:
s51: expanding operation H for the cavity judged at S4ep；
S52: obtaining void field HnpThe expression is as follows:
Hnp＝Hep-Ho
wherein HoIs a hollow region.
5. The method for RGB-D camera depth image restoration combined with color image according to claim 1, wherein the step S8 specifically includes: because the hole repaired by the S7 still has fine noise, in order to ensure the originality of the depth data as much as possible and remove the noise, the local filtering method is adopted to perform image filtering.
CN201810174885.5A 2018-03-02 2018-03-02 An RGB-D camera depth image inpainting method for joint color images Active CN108399632B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810174885.5A CN108399632B (en) 2018-03-02 2018-03-02 An RGB-D camera depth image inpainting method for joint color images

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810174885.5A CN108399632B (en) 2018-03-02 2018-03-02 An RGB-D camera depth image inpainting method for joint color images

Publications (2)

Publication Number Publication Date
CN108399632A CN108399632A (en) 2018-08-14
CN108399632B true CN108399632B (en) 2021-06-15

Family

ID=63091760

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810174885.5A Active CN108399632B (en) 2018-03-02 2018-03-02 An RGB-D camera depth image inpainting method for joint color images

Country Status (1)

Country Link
CN (1) CN108399632B (en)

Families Citing this family (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109345580B (en) * 2018-10-23 2020-03-24 北京字节跳动网络技术有限公司 Method and apparatus for processing image
CN109636732B (en) * 2018-10-24 2023-06-23 深圳先进技术研究院 Hole repairing method of depth image and image processing device
CN109360230A (en) * 2018-11-08 2019-02-19 武汉库柏特科技有限公司 A kind of method for registering images and system based on 2D camera Yu 3D camera
CN109712084B (en) * 2018-12-10 2021-01-19 上海奕瑞光电子科技股份有限公司 Image restoration method, image restoration system and flat panel detector
CN109685732B (en) * 2018-12-18 2023-02-17 重庆邮电大学 High-precision depth image restoration method based on boundary capture
CN109741405B (en) * 2019-01-21 2021-02-02 同济大学 Depth information acquisition system based on dual structured light RGB-D camera
CN111626086A (en) * 2019-02-28 2020-09-04 北京市商汤科技开发有限公司 Living body detection method, living body detection device, living body detection system, electronic device, and storage medium
CN111754713B (en) * 2019-03-28 2021-12-14 杭州海康威视数字技术股份有限公司 Video monitoring method, device and system
CN111862139B (en) * 2019-08-16 2023-08-18 中山大学 Dynamic object parametric modeling method based on color-depth camera
CN110866882B (en) * 2019-11-21 2021-09-07 湖南工程学院 Hierarchical joint bilateral filtering depth map inpainting method based on depth confidence
CN111524193B (en) * 2020-04-17 2022-05-03 西安交通大学 Method and device for measuring two-dimensional size of an object
CN111739031B (en) * 2020-06-19 2023-09-26 华南农业大学 Crop canopy segmentation method based on depth information
CN112116602A (en) * 2020-08-31 2020-12-22 北京的卢深视科技有限公司 Depth map repairing method and device and readable storage medium
CN112381867B (en) * 2020-11-09 2023-09-05 华南理工大学 Large-area depth image hole automatic filling method for industrial sorting pipeline
CN112991193B (en) * 2020-11-16 2022-09-23 武汉科技大学 Deep image restoration method, device and computer-readable storage medium
CN113408344B (en) * 2021-05-13 2024-11-26 深圳市捷顺科技实业股份有限公司 A method for generating three-dimensional face recognition and related device
CN113628205B (en) * 2021-08-25 2022-05-20 四川大学 Non-contact respiratory frequency detection method based on depth image
CN117541510B (en) * 2023-12-01 2024-11-01 维悟光子(北京)科技有限公司 Image restoration and completion method

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101067557A (en) * 2007-07-03 2007-11-07 北京控制工程研究所 Environment sensing one-eye visual navigating method adapted to self-aid moving vehicle
CN101374242A (en) * 2008-07-29 2009-02-25 宁波大学 A Depth Image Coding and Compression Method Applied to 3DTV and FTV Systems
CN101609508A (en) * 2008-06-18 2009-12-23 中国科学院自动化研究所 Sign structure and recognition methods to object identification and orientation information calculating
CN101662694A (en) * 2008-08-29 2010-03-03 深圳华为通信技术有限公司 Method and device for presenting, sending and receiving video and communication system
CN103217111A (en) * 2012-11-28 2013-07-24 西南交通大学 Non-contact contact line geometrical parameter detecting method
CN103561258A (en) * 2013-09-25 2014-02-05 同济大学 Kinect depth video spatio-temporal union restoration method
CN104571482A (en) * 2013-10-22 2015-04-29 中国传媒大学 Digital device control method based on somatosensory recognition
CN107622480A (en) * 2017-09-25 2018-01-23 长春理工大学 A Kinect depth image enhancement method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101067557A (en) * 2007-07-03 2007-11-07 北京控制工程研究所 Environment sensing one-eye visual navigating method adapted to self-aid moving vehicle
CN101609508A (en) * 2008-06-18 2009-12-23 中国科学院自动化研究所 Sign structure and recognition methods to object identification and orientation information calculating
CN101374242A (en) * 2008-07-29 2009-02-25 宁波大学 A Depth Image Coding and Compression Method Applied to 3DTV and FTV Systems
CN101662694A (en) * 2008-08-29 2010-03-03 深圳华为通信技术有限公司 Method and device for presenting, sending and receiving video and communication system
CN103217111A (en) * 2012-11-28 2013-07-24 西南交通大学 Non-contact contact line geometrical parameter detecting method
CN103561258A (en) * 2013-09-25 2014-02-05 同济大学 Kinect depth video spatio-temporal union restoration method
CN104571482A (en) * 2013-10-22 2015-04-29 中国传媒大学 Digital device control method based on somatosensory recognition
CN107622480A (en) * 2017-09-25 2018-01-23 长春理工大学 A Kinect depth image enhancement method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
《一种深度图像修复算法研究》;刘田间等;《信息技术》;20171231(第6期);全文 *
《基于Kinect的深度图像修复方法》;吕朝辉等;《吉林大学学报》;20160930;第46卷(第5期);全文 *
《面向工业应用的机器人手眼标定与物体定位》;程玉立等;《中国优秀硕士学位论文全文数据库 信息科技辑》;20160815(第08期);全文 *

Also Published As

Publication number Publication date
CN108399632A (en) 2018-08-14

Similar Documents

Publication Publication Date Title
CN108399632B (en) An RGB-D camera depth image inpainting method for joint color images
Chen et al. Depth image enhancement for Kinect using region growing and bilateral filter
Zhang et al. Fast haze removal for nighttime image using maximum reflectance prior
Ancuti et al. Night-time dehazing by fusion
Liu et al. Guided inpainting and filtering for kinect depth maps
CN108564597B (en) Video foreground object extraction method fusing Gaussian mixture model and H-S optical flow method
CN104680496B (en) A kind of Kinect depth map restorative procedures based on color images
TWI526992B (en) Method for optimizing occlusion in augmented reality based on depth camera
CN105205794B (en) A kind of synchronous enhancing denoising method of low-light (level) image
CN110232389A (en) A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance
CN110428439A (en) A kind of shadow detection method based on shadow region color saturation property
CN108038887A (en) Based on binocular RGB-D camera depth profile methods of estimation
CN110717935B (en) Image matching method, device and system based on image characteristic information
Wang et al. Single-image dehazing using color attenuation prior based on haze-lines
KR102327304B1 (en) A method of improving the quality of 3D images acquired from RGB-depth camera
Wang et al. Review of single image defogging
CN110827209A (en) An adaptive depth image inpainting method combining color and depth information
CN116263942A (en) Method for adjusting image contrast, storage medium and computer program product
CN109410160A (en) The infrared polarization image interfusion method driven based on multiple features and feature difference
CN106408533A (en) Card image extraction method and card image extraction system
Tran et al. Single Image Dehazing via Regional Saturation-Value Translation
Zhang et al. Nighttime haze removal with illumination correction
CN109636822A (en) A kind of improvement Canny adaptive edge extracting method based on new building subordinating degree function
CN108765477A (en) A kind of Kinect depth map restorative procedures based on optimization
US20210241430A1 (en) Methods, devices, and computer program products for improved 3d mesh texturing

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant