WO2017156905A1 - 一种将二维图像转化为多视点图像的显示方法及系统 - Google Patents

一种将二维图像转化为多视点图像的显示方法及系统 Download PDF

Info

Publication number
WO2017156905A1
WO2017156905A1 PCT/CN2016/086466 CN2016086466W WO2017156905A1 WO 2017156905 A1 WO2017156905 A1 WO 2017156905A1 CN 2016086466 W CN2016086466 W CN 2016086466W WO 2017156905 A1 WO2017156905 A1 WO 2017156905A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
viewpoint
depth value
display
view
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.)
Ceased
Application number
PCT/CN2016/086466
Other languages
English (en)
French (fr)
Inventor
徐遥令
侯志龙
姜晓飞
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.)
Shenzhen Skyworth RGB Electronics Co Ltd
Original Assignee
Shenzhen Skyworth RGB Electronics Co Ltd
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 Shenzhen Skyworth RGB Electronics Co Ltd filed Critical Shenzhen Skyworth RGB Electronics Co Ltd
Priority to AU2016397878A priority Critical patent/AU2016397878B2/en
Priority to US15/759,714 priority patent/US10334231B2/en
Publication of WO2017156905A1 publication Critical patent/WO2017156905A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/30—Image reproducers
    • H04N13/302—Image reproducers for viewing without the aid of special glasses, i.e. using autostereoscopic displays
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
    • H04N13/106—Processing image signals
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
    • H04N13/106—Processing image signals
    • H04N13/128—Adjusting depth or disparity
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/261—Image signal generators with monoscopic-to-stereoscopic image conversion
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/261—Image signal generators with monoscopic-to-stereoscopic image conversion
    • H04N13/268—Image signal generators with monoscopic-to-stereoscopic image conversion based on depth image-based rendering [DIBR]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/30—Image reproducers
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N2013/0074—Stereoscopic image analysis
    • H04N2013/0081—Depth or disparity estimation from stereoscopic image signals

Definitions

  • the present invention relates to the field of 3D display technologies, and in particular, to a display method and system for converting a two-dimensional image into a multi-view image.
  • Multi-viewpoints generally convert 3D view or view + depth of field content into multi-viewpoints for naked-eye 3D display, or by converting single-view (or 2D) content into multiple
  • the naked eye 3D display is performed after the viewpoint, and the 2 viewpoints or the viewpoint + depth of field also contain little content.
  • a single-viewpoint is converted into multi-viewpoint content for naked-eye 3D display.
  • the method for transferring single-view content to multi-view content includes three-dimensional information reconstruction based on multiple images, three-dimensional information reconstruction based on virtual camera, three-dimensional information reconstruction based on speckle information, and the like, which are complicated and difficult to implement. These methods have problems such as image cracking, distortion jitter, and the like.
  • the present invention aims to provide a display method and system for converting a two-dimensional image into a multi-view image, aiming at solving the single-view content transfer in the prior art.
  • the method of multi-view content is complicated, difficult to implement, and has defects of image cracking and distortion jitter.
  • a display method for converting a two-dimensional image into a multi-view image includes:
  • the object object in the two-dimensional image is acquired and identified, the depth value of the object object is calculated according to the frequency component of the object object, and the layered image is generated according to the depth value;
  • the layered image is viewed through different preset viewpoint positions, and the viewpoint image is calibrated. According to the depth value difference of the layered image, different interpolation algorithms are used to estimate the blank area pixels of the virtual viewpoint image and fill the blank area, and the layered image is removed.
  • the layer spacing forms a virtual viewpoint image, and the viewpoint image is scaled to form a single viewpoint image output;
  • the single-viewpoint image is sequentially stored, the single-viewpoint image is numbered, the cavity area of each single-viewpoint image is detected and filled, and the single-viewpoint image is subjected to abrupt region inspection and smoothing, and then the single-viewpoint images are integrated into one frame.
  • the composite image is converted into a logical electrical signal and sent to the naked eye 3D display for display.
  • the display method for converting a two-dimensional image into a multi-view image wherein after the pre-processed two-dimensional image is preprocessed, the object object in the two-dimensional image is acquired and identified, and the object is calculated according to the frequency component of the object object.
  • the depth value of the object is layered according to the depth value to generate a layered image, including:
  • Sharpening the processed two-dimensional image searching for the boundary and contour of the object in the two-dimensional image to identify the object, segmenting the object in the two-dimensional image and acquiring The obtained object object is identified;
  • the object object is layered according to the depth value of each object object according to the difference between the layer and the depth value, and the layer spacing is the depth value of the object object, and a layered image is generated.
  • the display method for converting a two-dimensional image into a multi-view image wherein the layered image is viewed through different preset viewpoint positions, the viewpoint image is calibrated, and different interpolation is adopted according to the depth value difference of the layered image.
  • the algorithm estimates the blank area pixels of the virtual view image and fills in the blank area, removes the layer spacing of the layered image to form a virtual view point image, and scales the view point image to form a single view point image output, which specifically includes:
  • the layered image is viewed through different preset viewpoint positions, and the actual viewpoint position and the virtual viewpoint position in the viewpoint image are respectively calibrated;
  • the actual viewpoint image and the virtual viewpoint image are respectively scaled to a predetermined scale, and a corresponding single viewpoint image is generated and output.
  • the display method for converting a two-dimensional image into a multi-viewpoint image wherein the single-viewpoint image is sequentially stored, the single-viewpoint image is numbered, the void region of each single-viewpoint image is detected, and a filling process is performed, and each single is processed Viewpoint image for mutation region check and After smoothing, integrating each single viewpoint image into a composite image specifically includes:
  • the single-viewpoint image is sequentially stored, and the single-viewpoint image is numbered as 1, 2, ..., N-viewpoint, and the actual viewpoint is marked, where N is a natural number;
  • the pixels of the N single-view images are staggered, and the N single-view images are integrated into a composite image corresponding to the physical pixels of the naked-eye 3D display.
  • the display method for converting a two-dimensional image into a multi-view image, wherein after the output processing is performed on the composite image, converting the composite image into a logical electrical signal and transmitting it to the naked-eye 3D display screen for display includes:
  • the processed composite image After performing frame rate conversion, gamma correction, signal amplitude adjustment, and color gamut format conversion processing on the composite image, the processed composite image is converted into a logical electrical signal and sent to the naked eye 3D display for display.
  • a display system for converting a two-dimensional image into a multi-view image comprising:
  • Image layering module which is used to preprocess the two-dimensional image to be processed, and obtain two The object object in the dimension image is identified, the depth value of the object object is calculated according to the frequency component of the object object, and the layered image is generated according to the size of the depth value;
  • the single-view image output module is configured to view the layered image through different preset viewpoint positions, calibrate the viewpoint image, and use different interpolation algorithms to estimate the blank area pixels of the virtual viewpoint image and fill according to the depth value difference of the layered image. a blank area, the layer spacing of the layered image is removed to form a virtual viewpoint image, and the viewpoint image is scaled to form a single viewpoint image output;
  • a composite image generation module for sequentially storing single-viewpoint images, numbering single-viewpoint images, detecting a void region of each single-viewpoint image and performing a filling process, and performing a mutation region check and smoothing process on each single-viewpoint image,
  • the single viewpoint image is integrated into a composite image
  • the display module is configured to perform output processing on the composite image, and convert the composite image into a logical electrical signal and send it to the naked eye 3D display for display.
  • the display system for converting a two-dimensional image into a multi-view image wherein the image layering module specifically includes:
  • the identification unit is configured to perform sharpening processing on the two-dimensional image to be processed, search and detect the boundary and contour of the object in the two-dimensional image to identify the object object, divide the object object in the two-dimensional image and acquire, and perform the acquired object object Identification
  • a depth value calculation unit configured to acquire a gray component of each object object, perform a Fourier transform on the gray component, obtain a frequency component of each object object, and calculate a depth value of each object object according to the frequency component;
  • a layering unit configured to correspond to a depth value according to a layer according to a depth value of each object object,
  • the layer spacing is the difference between the depth values of the object objects to layer the object objects to generate a layered image.
  • the display system for converting a two-dimensional image into a multi-view image wherein the single-view image output module specifically includes:
  • a viewpoint calibration unit configured to view the layered image through different preset viewpoint positions, and respectively calibrate the actual viewpoint position and the virtual viewpoint position in the viewpoint image;
  • the virtual view obtaining unit is configured to obtain a depth value of the object object where the calibrated virtual view point is located, perform interpolation calculation according to the difference of the depth value of the object object, and obtain a blank area of the calibrated virtual view according to the calculated result. Filling the pixels of each position to remove the layer spacing of the layered image to form a virtual viewpoint image;
  • the image output unit is configured to respectively scale the actual viewpoint image and the virtual viewpoint image to a predetermined scale to generate a corresponding single viewpoint image and output the image.
  • the display system for converting a two-dimensional image into a multi-view image wherein the composite image generating module specifically includes:
  • the actual viewpoint labeling unit is configured to sequentially store the single viewpoint images, and number the single viewpoint images as 1, 2, ..., N viewpoints, and mark the actual viewpoints, wherein N is a natural number;
  • a pixel interpolation unit for sequentially checking whether a single-view image other than the actual viewpoint has a void area, and if there is a void area, using a non-void area around the void area and four pixels adjacent to the hollow area, according to the distance Pixel interpolation and filling according to a certain specific gravity;
  • the mutation region processing unit is configured to sequentially check whether the single-view image other than the actual viewpoint has a mutation region, and if there is a mutation region, extract the position marker of the mutation region, and check the position region corresponding to the same position mark of the actual viewpoint image. Or the adjacent area is If there is a region with the same mutation, if the same viewpoint region exists in the actual viewpoint image, the detected single-viewpoint image mutation region is normal, otherwise the noise reduction smoothing process is performed;
  • the composite image generating unit is configured to interleave the pixels of the N single-view images according to the physical pixel arrangement of the naked-eye 3D display, and integrate the N single-view images into one physical pixel with the naked-eye 3D display. Corresponding composite image.
  • the display system for converting a two-dimensional image into a multi-view image wherein the display module specifically includes:
  • the image processing and display unit is configured to perform frame rate conversion, gamma correction, signal amplitude adjustment, and color gamut format conversion processing on the composite image, and convert the processed composite image into a logical electrical signal, and send the image to a naked eye 3D display. display.
  • the present invention provides a display method and system for converting a two-dimensional image into a multi-view image.
  • the object object is acquired and identified, and the depth value is estimated according to the frequency component of the object object, and according to the depth value.
  • the size of the object is layered to form a virtual object layered image space, and the virtual view image area is calibrated, and according to the depth value difference of the layered image, after filling with different interpolation algorithms, a single view image output is formed, and each view point is formed.
  • a composite image of multiple viewpoints is generated, and the image is output processed and sent to the naked eye 3D display for display.
  • the invention can realize two-dimensional image to multi-viewpoint image and perform naked-eye 3D display, reduce image distortion, and is simple and convenient, and has low cost.
  • FIG. 1 is a flow chart of a preferred embodiment of a display method for converting a two-dimensional image into a multi-view image according to the present invention
  • FIG. 2a is a schematic diagram of an original two-dimensional image of a specific application embodiment of a display method for converting a two-dimensional image into a multi-view image according to the present invention
  • FIG. 2b is a schematic diagram of a hierarchical space after processing the original two-dimensional image in FIG. 2a in a specific application embodiment of a display method for converting a two-dimensional image into a multi-view image according to the present invention
  • FIG. 3a is a schematic diagram of a view of a specific application embodiment of a display method for converting a two-dimensional image into a multi-view image according to the present invention
  • FIG. 3b is a schematic diagram of a virtual application point of a specific application embodiment of a method for converting a two-dimensional image into a multi-view image according to the present invention
  • FIG. 4 is a schematic diagram of a pixel relationship between a pixel in a view image hole region and an adjacent non-void region in a specific application embodiment of a method for converting a two-dimensional image into a multi-view image according to the present invention
  • FIG. 5 is a functional block diagram of a preferred embodiment of a display system for converting a two-dimensional image into a multi-view image according to the present invention.
  • the present invention also provides a flowchart of a preferred embodiment of a method for converting a two-dimensional image into a multi-view image, as shown in FIG. 1, wherein the method includes:
  • Step S100 After preprocessing the two-dimensional image to be processed, acquiring an object object in the two-dimensional image and identifying, and calculating a depth of the object object according to the frequency component of the object object The value is layered according to the depth value to generate a layered image.
  • the input two-dimensional image (referred to as the original two-dimensional image) is received, the two-dimensional image is sharpened, and the like, and the object object in the two-dimensional image is segmented and acquired as OBJi; and then according to the object object OBJi
  • the frequency component is used to estimate the depth value DP_OBJi; finally, the object layering is performed according to the size of the depth value DP_OBJi, and each physical object OBJi is placed in a different layer to form a virtual hierarchical space.
  • step S100 specifically includes:
  • Step S101 performing a sharpening process on the two-dimensional image to be processed, searching and detecting the boundary of the object in the two-dimensional image and the contour recognition object object, dividing the object object in the two-dimensional image and acquiring, and identifying the acquired object object;
  • Step S102 acquiring gray components of each object object, performing Fourier transform on the gray component, obtaining frequency components of the respective object objects, and calculating depth values of the respective object objects according to the frequency components;
  • Step S103 Layer the object object according to the depth value of each object object according to the difference between the layer and the depth value and the layer spacing as the depth value of the object object, to generate a layered image.
  • the object acquisition is first performed.
  • the original two-dimensional image is sharpened to make the edge and outline of the object in the image protrude, and the image is further divided by searching and detecting the boundary and contour of the object in the image, and identifying the object object.
  • the acquired object object is identified as OBJi, i takes 1, 2, ...; OBJi includes object object image content, object position information, and the like.
  • each object object OBJi (i takes 1, 2, ...;) is grayed out, the gray component of each object is obtained, and then the gray component is Fu.
  • the frequency component of each object is obtained after the transformation of the first leaf, and finally the depth value DP_OBJi of each object object is estimated according to the frequency distribution.
  • the depth value of each object object is estimated by the depth value calculation formula.
  • the object layering is performed according to its depth value DP_OBJi.
  • the DP_OBJi value is placed on the outer layer, the DP_OBJi value is small, and the inner layer is set.
  • the layer corresponds to the depth value, and the layer spacing is the difference between the depth values of the object object. , forming a virtual object layered space, imaginary distribution of object objects in the three-dimensional space.
  • the actual viewpoint in FIG. 2a is the original two-dimensional image
  • the object objects OBJ1 and OBJ2 are obtained after the object is acquired, and the depth values after the depth estimation are DP_OBJ1 and DP_OBJ2, respectively, and the object layer is layered.
  • the processed distribution is shown in Fig. 2b: OBJ1 and OBJ2 are specifically shown in Fig. 2a and Fig. 2b, and the Z direction is the depth direction.
  • Step S200 viewing a layered image by using different preset viewpoint positions, and calibrating the viewpoint image, and estimating a blank area pixel of the virtual viewpoint image by using different interpolation algorithms according to the difference of the depth value of the layered image, and filling the blank area, and removing the point.
  • the layer spacing of the layer images forms a virtual viewpoint image, and the viewpoint image is scaled to form a single viewpoint image output.
  • the layered image is viewed at different viewpoint positions, the viewpoint image is calibrated, and the virtual viewpoint image area is calibrated; and then the virtual viewpoint image is estimated by using different interpolation algorithms according to the difference of the depth values.
  • the pixels in the blank area are filled with blank areas, and then the layer spacing is removed to form a two-dimensional image, which forms a virtual viewpoint image. Image; and zooming the viewpoint image to form a single viewpoint image output.
  • step S200 specifically includes:
  • Step S201 viewing the layered image through different preset viewpoint positions, and respectively calibrating the actual viewpoint position and the virtual viewpoint position in the viewpoint image;
  • Step S202 Obtain a depth value of the object object where the calibrated virtual viewpoint is located, perform interpolation calculation according to the difference of the depth value of the object object, and obtain each position of the blank area of the calibrated virtual viewpoint according to the calculated result. After the pixels are filled, the layer spacing of the layered images is removed to form a virtual viewpoint image;
  • Step S203 The actual viewpoint image and the virtual viewpoint image are respectively scaled to a predetermined scale, and then a corresponding single viewpoint image is generated and output.
  • the viewpoint calibration is first performed. After receiving the layered image, set different viewpoint positions to view the layered image, and calibrate the viewpoint image; when the viewpoint image is viewed at a certain viewpoint position, and the formed viewpoint image is just the original two-dimensional image, the position is actual The position of the viewpoint, the original two-dimensional image is the actual single-view image; moving to the virtual viewpoint position on the same plane of the actual viewpoint position, to view the layered image, you can see different parts of the layered image object, and acquire the object of the layer object If there is a part that extends the boundary of the part, the virtual viewpoint is marked.
  • the layered image is viewed at the actual viewpoint position, and the formed viewpoint image is just the original single viewpoint image, and the original single viewpoint image is the actual viewpoint image.
  • View the layered image at the virtual viewpoint 1 position you can only see the existing part of OBJ2 (such as the left part of the OBJ2 dotted line), then get the left part of OBJ2; in addition to seeing all the existing OBJ1, you can actually see Not originally
  • the part (the white part of OBJ1) acquires all the image parts of OBJ1 and extends the boundary to the original part, thereby calibrating the virtual viewpoint 1, and the calibrated virtual viewpoint 1 is as shown in Fig. 3b.
  • the viewpoint interpolation is performed. If it is an actual viewpoint image, it will go directly to the next step without interpolation of the viewpoint; if it is a calibrated virtual viewpoint, the pixel of the blank area of the virtual viewpoint image is estimated and filled with blank according to the difference of the depth values, using different interpolation algorithms. The area forms a virtual viewpoint.
  • the viewpoint interpolation process is:
  • K4 S(m+1,n)/[S(m,n-1)+S(m,n+1)+S(m-1,n)+S(m+1,n)];
  • the pixel to be interpolated P(m,n) K1*P(m,n-1)+K2*P(m,n+1)+K3*P(m-1,n)+K4*P(m+1 , n).
  • the pixels of each position of the blank area of the calibrated virtual viewpoint are estimated by the interpolation algorithm, and filled, and then the layer spacing is removed to form a two-dimensional image, that is, a virtual viewpoint image is formed.
  • the viewpoint zoom is performed. Reduce or enlarge the viewpoint image (including the actual viewpoint image or the virtual viewpoint image) to a certain scale so as to meet the requirements of subsequent processing, for example, the viewpoint image is a 1920*1080 pixel matrix, and needs to be reduced to a 1366*768 pixel matrix, or The viewpoint image is a 1366*768 pixel matrix and needs to be enlarged to a 1920*1080 pixel matrix for subsequent processing.
  • a single viewpoint image output is formed.
  • Step S300 sequentially storing the single-viewpoint images, numbering the single-viewpoint images, detecting the void regions of the single-viewpoint images, performing the filling process, and performing the mutation region check and smoothing processing on the single-viewpoint images, and then setting the single-viewpoint image sets. Become a composite image.
  • the single-viewpoint image is sequentially stored, and the single-viewpoint image is numbered as 1, 2, ..., N-viewpoints, and the cavity region inspection and filling process of each viewpoint image is performed, and each is performed.
  • the mutation region inspection and smoothing processing of the viewpoint image integrates the N viewpoint images into one composite image to complete the viewpoint synthesis.
  • step S300 specifically includes:
  • Step S301 sequentially storing single-viewpoint images, and numbering the single-viewpoint images as 1, 2, ..., N viewpoints, and marking the actual viewpoints, where N is a natural number;
  • Step S302 sequentially check whether there is a void area other than the actual viewpoint, and if there is a void area, use the non-void area around the void area and the four pixels adjacent to the hollow area, according to the distance and the distance The specific gravity of the pixel interpolation and filling;
  • Step S303 sequentially checking whether there is a region in which the single-view image other than the actual viewpoint has a mutation, and if there is a region with a mutation, extracting the position marker of the abrupt region, and checking whether the position region or the adjacent region corresponding to the same position mark of the actual viewpoint image is If there is a region of the same mutation, if the same viewpoint region exists in the actual viewpoint image, the detected single-viewpoint image mutation region is normal, otherwise the noise reduction smoothing process is performed;
  • Step S304 according to the physical pixel arrangement of the naked-eye 3D display screen, the pixels of the N single-viewpoint images are staggered, and the N single-viewpoint images are integrated into a composite image corresponding to the physical pixels of the naked-eye 3D display screen. .
  • the viewpoint fusion includes viewpoint storage, hole region filling, image abrupt region smoothing, and viewpoint integration.
  • a single-viewpoint image is received, and viewpoint storage is performed: the single-viewpoint image is sequentially stored, and the single-viewpoint image storage is numbered 1, 2, ..., N-viewpoint, and the actual viewpoint is simultaneously marked.
  • the hole region filling of the viewpoint image is performed: sequentially checking other single viewpoint images other than the actual viewpoint, that is, checking whether the image has a void region (all black pixel regions), and if there is a void region: using a non-hollow region around the void region Adjacent to the void area
  • the four pixels are pixel-interpolated and filled according to the distance from the distance, that is, the pixel to be filled in the cavity area P(x, y), and the pixels in the non-void area around the cavity area are P(x1, y), P, respectively.
  • the black area is the void area, One pixel P(x, y) to be filled, four pixels P(x1, y), P(x2, y), P(x, y1), P(x, y2) of the non-void area around the cavity area Adjacent to the cavity area, P(x1, y) and P(x2, y) have the same y-coordinate
  • the viewpoint integration is performed: according to the physical pixel arrangement of the naked-eye 3D display screen, the pixels of the N single-viewpoint images are staggered, and the N single-viewpoint images are integrated into one composite image, and each pixel of the composite image and the 3D screen are synthesized.
  • One-to-one correspondence of physical pixels including pixels of each single-view image and physical pixels corresponding to the view area of the 3D screen A correspondence.
  • Step S400 After performing output processing on the composite image, converting the composite image into a logical electrical signal and transmitting it to the naked eye 3D display screen for display.
  • the composite image is woven into a logical electrical signal (such as LVDS signal, VBO signal, TTL signal, etc.) and sent to the naked eye.
  • a logical electrical signal such as LVDS signal, VBO signal, TTL signal, etc.
  • the naked-eye 3D display receives the composite image to achieve naked-eye 3D display.
  • the present invention proposes to convert a two-dimensional image into a display of a multi-view image: sharpening the input two-dimensional image, dividing and acquiring the object object in the two-dimensional image and identifying it as OBJi, Further, the depth value DP_OBJi is estimated according to the frequency component of the object object OBJi, and the object layering is performed according to the size of the depth value DP_OBJi to form a virtual object layered image space; then the layered image is viewed at different viewpoint positions, and the calibration is performed.
  • the virtual view image area further estimates the blank area pixel and the filled blank area of the virtual view image according to the difference of the depth value of the layered image, further removes the layer interval to form a virtual view point image, and zooms the view point image, Forming a single-viewpoint image output; finally storing the single-viewpoint image sequentially, and numbering the single-viewpoint image as 1, 2, ..., N-viewpoints, further performing void region inspection and filling processing of each viewpoint image, and performing image processing of each viewpoint image Mutation area check and smoothing, and N viewpoint images
  • the composite image is woven into a logical electrical signal and sent to the naked eye 3D screen, and the naked eye 3D screen receives the composite image signal to realize the naked eye 3D. display.
  • the invention can realize two-dimensional image transfer to multi-viewpoint image and carry out The naked eye 3D display, as well as reducing image distortion, is simple, convenient, low cost, and has
  • the present invention also provides a functional block diagram of a preferred embodiment of a display system for converting a two-dimensional image into a multi-view image, as shown in FIG. 5, wherein the system includes:
  • the image layering module 100 is configured to: after the pre-processing of the two-dimensional image to be processed, acquire an object object in the two-dimensional image and identify the depth value of the object object according to the frequency component of the object object, and divide the depth value according to the depth value.
  • the layer generates a layered image; as described above in the method embodiments.
  • the single-view image output module 200 is configured to view the layered image through different preset viewpoint positions, calibrate the viewpoint image, and estimate the blank area pixels of the virtual viewpoint image by using different interpolation algorithms according to the depth value difference of the layered image.
  • the blank area is filled, the layer spacing of the layered image is removed to form a virtual viewpoint image, and the viewpoint image is scaled to form a single viewpoint image output; as described in the above method embodiment.
  • the composite image generation module 300 is configured to sequentially store the single-viewpoint images, number the single-viewpoint images, detect the void regions of the single-viewpoint images, perform the filling process, and perform the mutation region check and smoothing processing on the single-viewpoint images, and then Each single viewpoint image is integrated into a composite image; as described above in the method embodiments.
  • the display module 400 is configured to perform the output processing on the composite image, and then convert the composite image into a logical electrical signal and send it to the naked eye 3D display screen for display; as described in the foregoing method embodiment.
  • the display system for converting a two-dimensional image into a multi-view image wherein the image layering module specifically includes:
  • the identification unit is configured to perform sharpening processing on the two-dimensional image to be processed, search and detect the boundary and contour of the object in the two-dimensional image to identify the object object, divide the object object in the two-dimensional image and acquire, and perform the acquired object object Identification; as described above in the method embodiments.
  • a depth value calculation unit configured to obtain a gray component of each object object, perform a Fourier transform on the gray component, and obtain a frequency component of each object object, and calculate a depth value of each object object according to the frequency component; Said.
  • the layering unit is configured to layer the object object according to the depth value of each object object according to the difference between the layer and the depth value and the layer spacing as the depth value of the object object, to generate a layered image; as described in the foregoing method embodiment.
  • the display system for converting a two-dimensional image into a multi-view image wherein the single-view image output module specifically includes:
  • the viewpoint calibration unit is configured to view the layered image through different preset viewpoint positions, and respectively calibrate the actual viewpoint position and the virtual viewpoint position in the viewpoint image; as described in the above method embodiment.
  • the virtual view obtaining unit is configured to obtain a depth value of the object object where the calibrated virtual view point is located, perform interpolation calculation according to the difference of the depth value of the object object, and obtain a blank area of the calibrated virtual view according to the calculated result.
  • the pixels of each position are padded, and the layer spacing of the layered image is removed to form a virtual view image; as described in the above method embodiment.
  • An image output unit configured to respectively generate an actual single viewpoint image and a virtual viewpoint image to a predetermined scale, and generate a corresponding single viewpoint image and output the same; Said.
  • the display system for converting a two-dimensional image into a multi-view image wherein the composite image generating module specifically includes:
  • the actual viewpoint labeling unit is configured to sequentially store the single viewpoint images, and number the single viewpoint images as 1, 2, ..., N viewpoints, and mark the actual viewpoints, where N is a natural number; as described in the above method embodiment.
  • a pixel interpolation unit for sequentially checking whether a single-view image other than the actual viewpoint has a void area, and if there is a void area, using a non-void area around the void area and four pixels adjacent to the hollow area, according to the distance
  • the pixel is interpolated and filled according to a certain specific gravity; as described in the above method embodiment.
  • the mutation region processing unit is configured to sequentially check whether the single-view image other than the actual viewpoint has a mutation region, and if there is a mutation region, extract the position marker of the mutation region, and check the position region corresponding to the same position mark of the actual viewpoint image. Or whether there is a region of the same mutation in the adjacent region. If the same viewpoint region exists in the actual viewpoint image, the examined single-viewpoint image mutation region is normal, otherwise the noise reduction smoothing process is performed; as described in the above method embodiment.
  • the composite image generating unit is configured to interleave the pixels of the N single-view images according to the physical pixel arrangement of the naked-eye 3D display, and integrate the N single-view images into one physical pixel with the naked-eye 3D display. Corresponding composite image; as described in the method embodiment above.
  • the display system for converting a two-dimensional image into a multi-view image wherein the display module specifically includes:
  • the image processing and display unit is configured to perform frame rate conversion, gamma correction, signal amplitude adjustment, and color gamut format conversion processing on the composite image, and convert the processed composite image into a logical electrical signal, and send the image to a naked eye 3D display. Display; as described above in the method embodiments.
  • the present invention provides a display method and system for converting a two-dimensional image into a multi-view image, the method comprising: acquiring an object object in a two-dimensional image to be processed and identifying, and calculating according to a frequency component of the object object
  • the depth value of the object object is hierarchically generated according to the depth value to generate a layered image; the layered image is viewed through different preset viewpoint positions, the viewpoint image is calibrated, and the virtual viewpoint image is estimated according to the depth value difference of the layered image.
  • Blank area pixels fill the blank area, generate single-view image output; store single-view image in turn, detect the void area of each single-view image and fill it, check the abrupt area and smooth it, and integrate each single-view image into one
  • the image is composited; the composite image is processed and sent to the naked eye 3D display.
  • the invention can realize two-dimensional image to multi-viewpoint image and perform naked-eye 3D display, reduce image distortion, and is simple and convenient, and has low cost.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)
  • Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)

Abstract

本发明公开了一种将二维图像转化为多视点图像的显示方法及系统，方法包括：获取待处理的二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，根据深度值大小进行分层生成分层图像；通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，估算出虚拟视点图像的空白区域像素并填充空白区域，生成单视点图像输出；依次存储单视点图像，检测各个单视点图像的空洞区域检查并进行填充处理、检查突变区域后进行平滑处理，各个单视点图像集成为一幅合成图像；将合成图像处理后发送至裸眼3D显示屏显示。本发明能够实现二维图像转多视点图像、及进行裸眼3D显示，减少图像失真，简单方便、成本低。

Description

一种将二维图像转化为多视点图像的显示方法及系统 技术领域
本发明涉及3D显示技术领域，尤其涉及一种将二维图像转化为多视点图像的显示方法及系统。
背景技术
裸眼3D电视的观察者可以不佩戴眼镜来进行3D显示体验。但现有技术中通常需要从不同的角度来体验裸眼3D电视的3D显示效果，裸眼3D电视需要呈现出略有差异的多个视点的图像。但现有的多视点的技术不成熟，多视点一般是通过将2视点或视点+景深的内容转换成多视点后进行裸眼3D显示、或是通过将单视点(或二维)内容转换成多视点后进行裸眼3D显示，2视点或视点+景深也包含内容很少。通常是将但单视点转换成多视点内容后进行裸眼3D显示。目前单视点内容转多视点内容的方法包括基于多幅图像的三维信息重构、基于虚拟摄像机的三维信息重构、基于散斑信息的三维信息重构等方法，这些方法复杂、实现难度大，这些方法存在图像裂缝、失真抖动等问题。
因此，现有技术还有待于改进和发展。
发明内容
鉴于现有技术的不足，本发明目的在于提供一种将二维图像转化为多视点图像的显示方法及系统，旨在解决现有技术中单视点内容转 多视点内容的方法，方法复杂、实现难度大，存在图像裂缝、失真抖动的缺陷。
本发明的技术方案如下：
一种将二维图像转化为多视点图像的显示方法，其中，方法包括；
对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像；
通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出；
依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像；
对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示。
所述的将二维图像转化为多视点图像的显示方法，其中，所述对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像具体包括：
对待处理的二维图像进行锐化处理，搜索检测二维图像中的物体的边界和轮廓识别物体对象，分割二维图像中的物体对象并获取，将 获取到的物体对象进行标识；
获取各个物体对象的灰度分量，对灰度分量进行傅立叶变换后得到各个物体对象的频率分量，根据频率分量计算出各个物体对象的深度值；
根据各个物体对象的深度值按照层与深度值对应、层间距为物体对象的深度值之差对物体对象进行分层，生成分层图像。
所述的将二维图像转化为多视点图像的显示方法，其中，所述通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出具体包括：
通过预先设定的不同的视点位置观看分层图像，分别标定出视点图像中的实际视点位置和虚拟视点位置；
获取标定的虚拟视点所在的物体对象的深度值，根据物体对象的深度值的差异采用不同的插值算法进行插值计算，根据计算后的结果获取标定的虚拟视点的空白区域的各个位置的像素后进行填充，去掉分层图像的层间距形成虚拟视点图像；
将实际视点图像和虚拟视点图像分别缩放至预定的尺度后生成对应的单视点图像并输出。
所述的将二维图像转化为多视点图像的显示方法，其中，所述依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和 平滑处理后，将各个单视点图像集成为一幅合成图像具体包括：
依次存储单视点图像，并将单视点图像编号为1、2、...、N视点，并标注出实际视点，其中N为自然数；
依次检查除实际视点外的其它单视点图像是否存在空洞区域，若存在空洞区域，则采用空洞区域周围的非空洞区域且与空洞区域相邻的四个像素、按照距离的远近按一定的比重进行像素插值和填充；
依次检查除实际视点外的其它单视点图像是否有突变的区域，若存在突变的区域，提取突变区域的位置标记，并检查实际视点图像的相同位置标记对应的位置区域或临近区域是否存在同样的突变的区域，若实际视点图像存在同样的突变区域，则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理；
根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅与裸眼3D显示屏的物理像素一一对应的合成图像。
所述的将二维图像转化为多视点图像的显示方法，其中，所述对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示具体包括：
对合成图像进行帧频转换、Gamma校正、信号幅度调整、色域格式转换处理后，将处理后的合成图像转化成逻辑电信号，并发送至裸眼3D显示屏进行显示。
一种将二维图像转化为多视点图像的显示系统，其中，系统包括：
图像分层模块，用于对待处理的二维图像进行预处理后，获取二 维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像；
单视点图像输出模块，用于通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出；
合成图像生成模块，用于依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像；
显示模块，用于对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示。
所述的将二维图像转化为多视点图像的显示系统，其中，所述图像分层模块具体包括：
标识单元，用于对待处理的二维图像进行锐化处理，搜索检测二维图像中的物体的边界和轮廓识别物体对象，分割二维图像中的物体对象并获取，将获取到的物体对象进行标识；
深度值计算单元，用于获取各个物体对象的灰度分量，对灰度分量进行傅立叶变换后得到各个物体对象的频率分量，根据频率分量计算出各个物体对象的深度值；
分层单元，用于根据各个物体对象的深度值按照层与深度值对应、 层间距为物体对象的深度值之差对物体对象进行分层，生成分层图像。
所述的将二维图像转化为多视点图像的显示系统，其中，所述单视点图像输出模块具体包括：
视点标定单元，用于通过预先设定的不同的视点位置观看分层图像，分别标定出视点图像中的实际视点位置和虚拟视点位置；
虚拟视点获取单元，用于获取标定的虚拟视点所在的物体对象的深度值，根据物体对象的深度值的差异采用不同的插值算法进行插值计算，根据计算后的结果获取标定的虚拟视点的空白区域的各个位置的像素后进行填充，去掉分层图像的层间距形成虚拟视点图像；
图像输出单元，用于将实际视点图像和虚拟视点图像分别缩放至预定的尺度后生成对应的单视点图像并输出。
所述的将二维图像转化为多视点图像的显示系统，其中，所述合成图像生成模块具体包括：
实际视点标注单元，用于依次存储单视点图像，并将单视点图像编号为1、2、...、N视点，并标注出实际视点，其中N为自然数；
像素插值单元，用于依次检查除实际视点外的其它单视点图像是否存在空洞区域，若存在空洞区域，则采用空洞区域周围的非空洞区域且与空洞区域相邻的四个像素、按照距离的远近按一定的比重进行像素插值和填充；
突变区域处理单元，用于依次检查除实际视点外的其它单视点图像是否有突变的区域，若存在突变的区域，提取突变区域的位置标记，并检查实际视点图像的相同位置标记对应的位置区域或临近区域是 否存在同样的突变的区域，若实际视点图像存在同样的突变区域，则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理；
合成图像生成单元，用于根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅与裸眼3D显示屏的物理像素一一对应的合成图像。
所述的将二维图像转化为多视点图像的显示系统，其中，所述显示模块具体包括：
图像处理及显示单元，用于对合成图像进行帧频转换、Gamma校正、信号幅度调整、色域格式转换处理后，将处理后的合成图像转化成逻辑电信号，并发送至裸眼3D显示屏进行显示。
本发明提供了一种将二维图像转化为多视点图像的显示方法及系统，通过对二维图像预处理后，获取物体对象并标识，根据物体对象的频率分量估算深度值，并根据深度值大小进行对象分层，形成虚拟的对象分层图像空间，并标定虚拟视点图像区域，根据分层图像的深度值差异、采用不同的插值算法进行填充后，形成单视点图像输出，并对各视点图像进行平滑处理后，生成多视点的合成图像，并将图像进行输出处理后发送至裸眼3D显示屏进行显示。本发明能够实现二维图像转多视点图像、及进行裸眼3D显示，减少图像失真，简单方便、成本低。
附图说明
图1为本发明中一种将二维图像转化为多视点图像的显示方法的较佳实施例的流程图；
图2a为本发明中一种将二维图像转化为多视点图像的显示方法的具体应用实施例原始二维图像示意图；
图2b为本发明中一种将二维图像转化为多视点图像的显示方法的具体应用实施例中对图2a中的原始二维图像处理后分层空间示意图；
图3a为本发明中一种将二维图像转化为多视点图像的显示方法的具体应用实施例视点标定示意图；
图3b为本发明中一种将二维图像转化为多视点图像的显示方法的具体应用实施例虚拟视点示意图；
图4为本发明中一种将二维图像转化为多视点图像的显示方法的具体应用实施例视点图像空洞区域像素及相邻非空洞区域像素关系的示意图；
图5为本发明的一种将二维图像转化为多视点图像的显示系统的较佳实施例的功能原理框图。
具体实施方式
为使本发明的目的、技术方案及效果更加清楚、明确，以下对本发明进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
本发明还提供了一种将二维图像转化为多视点图像的显示方法的较佳实施例的流程图，如图1所示，其中，方法包括：
步骤S100、对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度 值，按照深度值的大小进行分层生成分层图像。
具体地，接收输入的二维图像(称为原始二维图像)，对二维图像进行锐化等处理，分割并获取到二维图像中的物体对象并标识为OBJi；然后根据物体对象OBJi的频率分量来估算出其深度值DP_OBJi；最后按照深度值DP_OBJi的大小进行对象分层，将各个物理对象OBJi放在不同的层、形成虚拟分层空间。
进一步实施时，所述步骤S100具体包括：
步骤S101、对待处理的二维图像进行锐化处理，搜索检测二维图像中的物体的边界和轮廓识别物体对象，分割二维图像中的物体对象并获取，将获取到的物体对象进行标识；
步骤S102、获取各个物体对象的灰度分量，对灰度分量进行傅立叶变换后得到各个物体对象的频率分量，根据频率分量计算出各个物体对象的深度值；
步骤S103、根据各个物体对象的深度值按照层与深度值对应、层间距为物体对象的深度值之差对物体对象进行分层，生成分层图像。
具体实施时，首先进行对象获取。接收到原始二维图像后，对原始二维图像进行锐化处理、使得图像中物体的边缘和轮廓线突出，通过搜索和检测图像中物体的边界和轮廓、识别到物体对象，进一步分割图像中的物体对象并获取到物体对象。将获取到的物体对象标识为OBJi，i取1、2、...；OBJi包括物体对象图像内容、对象位置信息等。
其次进行深度估算。首先对各个物体对象OBJi(i取1、2、...；)进行灰度化处理、获取各个对象的灰度分量，然后对灰度分量进行傅 立叶变换后得到各个对象的频率分量，最后根据频率分布来估算出各个物体对象的深度值DP_OBJi。物理对象OBJi的频率分量为OBJi_f，则其深度值计算公式为DP_OBJi＝(OBJi_f/frq_s-1)*DP_U+DP_S，其中frq_s为设定的参考基准频率、DP_S为设定的参考基准深度值。通过深度值计算公式估算出各个物体对象的深度值。
最后进行对象分层。物体对象分层按照其深度值DP_OBJi来进行，DP_OBJi值大则放在外层、DP_OBJi值小则放在内层，层与深度值对应、层间距为物体对象的深度值之差；通过对象分层，形成一个虚拟的对象分层空间，虚构出物体对象在立体空间的分布。
例如，如图2a、图2b所示：图2a中实际视点为原始二维图像；经过对象获取后得到物体对象OBJ1和OBJ2、以及经过深度估算后深度值分别为DP_OBJ1和DP_OBJ2，经过对象分层处理后的分布如图图2b所示：OBJ1、OBJ2具体如图2a和图2b所示，Z方向为深度方向。
步骤S200、通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出。
具体实施时，接收分层图像后，在不同的视点位置观看分层图像，进行视点图像标定、标定出虚拟视点图像区域；然后根据深度值的差异、采用不同的插值算法估算出虚拟视点图像的空白区域的像素、并填充空白区域，然后去掉层间距形成二维图像、即形成了虚拟视点图 像；以及对视点图像进行缩放，形成单视点图像输出。
具体地，步骤S200具体包括：
步骤S201、通过预先设定的不同的视点位置观看分层图像，分别标定出视点图像中的实际视点位置和虚拟视点位置；
步骤S202、获取标定的虚拟视点所在的物体对象的深度值，根据物体对象的深度值的差异采用不同的插值算法进行插值计算，根据计算后的结果获取标定的虚拟视点的空白区域的各个位置的像素后进行填充，去掉分层图像的层间距形成虚拟视点图像；
步骤S203、将实际视点图像和虚拟视点图像分别缩放至预定的尺度后生成对应的单视点图像并输出。
具体实施时，首先进行视点标定。接收分层图像后，设定不同的视点位置来观看分层图像，标定出视点图像；当某一视点位置观看分层图像时、形成的视点图像刚好是原始二维图像，则此位置为实际视点位置、原始二维图像是实际单视点图像；在实际视点位置的同一平面上移动至虚拟视点位置、来观看分层图像，可以看到分层图像物体对象的不同部分，获取物体对象的已有部分、并延伸未有部分的边界、则标定出虚拟视点。
例如，如图3a、图3b所示：图3a中，在实际视点位置观看分层图像，形成的视点图像刚好是原始单视点图像，则原始单视点图像为实际视点图像。在虚拟视点1位置观看分层图像：仅仅可以看到OBJ2的已有的一部分(如OBJ2虚线的左边部分)、则获取OBJ2的左边一部分；除了可以看到OBJ1已有的全部、实际还可以看到原来没有的 部分(OBJ1的白色部分)，则获取OBJ1已有全部图像部分、并进行边界延伸至原来没有的部分，从而标定出虚拟视点1，标定出的虚拟视点1如图3b所示。
其次进行视点插值。如果是实际视点图像，将不进行视点插值而直接进入下一步处理；是标定的虚拟视点，则根据深度值的差异、采用不同的插值算法估算出虚拟视点图像的空白区域的像素、并填充空白区域，形成虚拟视点。视点插值过程为：
首先对标定的虚拟视点的各层对象的深度值进行判断：
1)如果对象OBJi的深度值、DP_STD为设定阀值，则采用四点平均算法来进行插值，即：待插值像素P(m，n)周围相邻的四个像素分别为P(m，n-1)、P(m，n+1)、P(m-1，n)、P(m+1，n)，则插入的像素P(m，n)＝[P(m，n-1)+P(m，n+1)+P(m-1，n)+P(m+1，n)]/4，m、n为行列标记；如果相邻的四个像素中某个像素不存在、则用临近的行像素或列像素替代，例如P(m，n+1)和P(m+1，n)不存在，在插入的像素P(m，n)＝[P(m，n-1)+P(m，n-1)+P(m-1，n)+P(m-1，n)]/4。
2)如果对象OBJi的深度值、则根据相邻图像像素的变化趋势来进行插值，例如采用相邻八个像素点插值：待插值像素P(m，n)，其周围相邻的四个像素分别为P(m，n-2)、P(m，n-1)、P(m，n+1)、P(m，n+2)、P(m-2，n)、P(m-1，n)、P(m+1，n)、P(m+2，n)；设S(m，n-1)＝P(m，n-1)-P(m，n-2)、S(m，n+1)＝P(m，n+2)-P(m，n+1)、S(m-1，n)＝P(m-2，n)-P(m-1，n)、S(m+1，n)＝P(m+2，n)-P(m+1，n)，以及K1＝S(m，n-1)/[S(m，n-1)+S(m，n+1)+S(m-1，n)+S(m+1，n)]、K2＝S(m， n+1)/[S(m，n-1)+S(m，n+1)+S(m-1，n)+S(m+1，n)]、
K3＝S(m-1，n)/[S(m，n-1)+S(m，n+1)+S(m-1，n)+S(m+1，n)]、
K4＝S(m+1，n)/[S(m，n-1)+S(m，n+1)+S(m-1，n)+S(m+1，n)]；则待插值像素P(m，n)＝K1*P(m，n-1)+K2*P(m，n+1)+K3*P(m-1，n)+K4*P(m+1，n)。
通过插值算法估算出标定的虚拟视点的空白区域的各个位置的像素、并进行填充，然后去掉层间距、形成二维图像，即形成了虚拟视点图像。
最后进行视点缩放。将视点图像(包括实际视点图像或虚拟视点图像)缩小或放大至一定的尺度，使得符合后续处理的要求，例如：视点图像为1920*1080像素矩阵、需要缩小至1366*768像素矩阵，或是视点图像为1366*768像素矩阵、需要放大至1920*1080像素矩阵，以便后续处理。经过视点缩放后形成单视点图像输出。
步骤S300、依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像。
具体实施时，接收到单视点图像后，依次存储单视点图像、并将单视点图像编号为1、2、...、N视点，进行各个视点图像的空洞区域检查和填充处理、以及进行各个视点图像的突变区域检查和平滑处理，将N个视点图像集成为一幅合成图像，完成视点合成。
具体地，所述步骤S300具体包括：
步骤S301、依次存储单视点图像，并将单视点图像编号为1、2、...、N视点，并标注出实际视点，其中N为自然数；
步骤S302、依次检查除实际视点外的其它单视点图像是否存在空洞区域，若存在空洞区域，则采用空洞区域周围的非空洞区域且与空洞区域相邻的四个像素、按照距离的远近按一定的比重进行像素插值和填充；
步骤S303、依次检查除实际视点外的其它单视点图像是否有突变的区域，若存在突变的区域，提取突变区域的位置标记，并检查实际视点图像的相同位置标记对应的位置区域或临近区域是否存在同样的突变的区域，若实际视点图像存在同样的突变区域，则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理；
步骤S304、根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅与裸眼3D显示屏的物理像素一一对应的合成图像。
具体实施时，视点融合包括视点存储、空洞区域填充、图像突变区域平滑、视点集成。
首先接收到单视点图像，进行视点存储：依次存储单视点图像、并将单视点图像存储编号为1、2、...、N视点，以及同时将实际视点标注出来。
然后进行视点图像的空洞区域填充：依次检查除实际视点外的其它单视点图像、即检查图像是否有空洞区域(全部是黑色像素的区域)，如果有空洞区域：采用空洞区域周围的非空洞区域且与空洞区域相邻 的四个像素、按照距离的远近取比重进行像素插值和填充，即：空洞区域待填充像素P(x，y)，空洞区域周围的非空洞区域像素分别为P(x1，y)、P(x2，y)、P(x，y1)、P(x，y2)，此四个像素与空洞区域相邻、且与P(x，y)的距离分别为V1＝|x1-x|、V2＝|x2-x|、V3＝|y1-y|、V4＝|y2-y|，令V＝V1+V2+V3+V4，则P(x，y)＝P(x1，y)*V1/V+P(x2，y)*V2/V+P(x，y1)*V3/V+P(x，y2)*V4/V；例如，如图4所示：黑色区域为空洞区域、其待填充的一个像素P(x，y)，空洞区域周围的非空洞区域的四个像素P(x1，y)、P(x2，y)、P(x，y1)、P(x，y2)与空洞区域相邻，P(x1，y)和P(x2，y)与待填充像素P(x，y)具有相同的y坐标、水平间距分别为V1和V2，P(x，y1)和P(x，y2)与待填充像素P(x，y)具有相同的x坐标、水平间距分别为V3和V4；依次计算出每一个空洞区域的像素进行填充，消除存在的空洞。
进一步进行视点图像的突变区域平滑：依次检查除实际视点外的其它单视点图像，检查图像是否有突变的区域，如果单视点图像有图像突变区域：提取突变区域的位置标记PM，进一步检查实际视点图像的相同PM位置区域或临近区域是否存在同样的突变，如果实际视点图像存在同样的突变区域、则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理、防止图像突变。
最后进行视点集成：根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅合成图像，合成图像的每一个像素与3D屏的物理像素一一对应、包含的每一个单视点图像的像素与3D屏对应视点区域的物理像素一 一对应。
步骤S400、对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示。
具体实施时，对合成图像进行帧频转换、Gamma校正、信号幅度调整、色域格式转换等处理后，将合成图像编织成逻辑电信号(如LVDS信号、VBO信号、TTL信号等)发送给裸眼3D显示屏。裸眼3D显示屏接收合成图像，实现裸眼3D显示。
由以上方法实施例可知，本发明提出将二维图像转化为多视点图像的显示：对输入的二维图像进行锐化等处理、分割并获取到二维图像中的物体对象并标识为OBJi，进一步根据物体对象OBJi的频率分量来估算出其深度值DP_OBJi，以及按照深度值DP_OBJi的大小进行对象分层、形成虚拟的对象分层图像空间；然后在不同的视点位置观看分层图像、标定出虚拟视点图像区域，进一步根据分层图像的深度值差异、采用不同的插值算法估算出虚拟视点图像的空白区域像素及填充空白区域，进一步去掉层间距形成虚拟视点图像，以及对视点图像进行缩放、形成单视点图像输出；最后依次存储单视点图像、并将单视点图像编号为1、2、...、N视点，进一步进行各个视点图像的空洞区域检查和填充处理、并进行各个视点图像的突变区域检查和平滑处理、以及将N个视点图像集成为一幅合成图像，以及进一步对合成图像进行帧频转换和色域格式转换等输出处理后、将合成图像编织成逻辑电信号发送给裸眼3D屏，裸眼3D屏接收合成图像信号、实现裸眼3D显示。本发明能够实现二维图像转多视点图像、及进行 裸眼3D显示，以及减少图像失真，简单方便、成本低，具有较大应用价值。
本发明还提供了一种将二维图像转化为多视点图像的显示系统的较佳实施例的功能原理框图，如图5所示，其中，系统包括：
图像分层模块100，用于对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像；具体如上方法实施例所述。
单视点图像输出模块200，用于通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出；具体如上方法实施例所述。
合成图像生成模块300，用于依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像；具体如上方法实施例所述。
显示模块400，用于对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示；具体如上方法实施例所述。
所述的将二维图像转化为多视点图像的显示系统，其中，所述图像分层模块具体包括：
标识单元，用于对待处理的二维图像进行锐化处理，搜索检测二维图像中的物体的边界和轮廓识别物体对象，分割二维图像中的物体对象并获取，将获取到的物体对象进行标识；具体如上方法实施例所述。
深度值计算单元，用于获取各个物体对象的灰度分量，对灰度分量进行傅立叶变换后得到各个物体对象的频率分量，根据频率分量计算出各个物体对象的深度值；具体如上方法实施例所述。
分层单元，用于根据各个物体对象的深度值按照层与深度值对应、层间距为物体对象的深度值之差对物体对象进行分层，生成分层图像；具体如上方法实施例所述。
所述的将二维图像转化为多视点图像的显示系统，其中，所述单视点图像输出模块具体包括：
视点标定单元，用于通过预先设定的不同的视点位置观看分层图像，分别标定出视点图像中的实际视点位置和虚拟视点位置；具体如上方法实施例所述。
虚拟视点获取单元，用于获取标定的虚拟视点所在的物体对象的深度值，根据物体对象的深度值的差异采用不同的插值算法进行插值计算，根据计算后的结果获取标定的虚拟视点的空白区域的各个位置的像素后进行填充，去掉分层图像的层间距形成虚拟视点图像；具体如上方法实施例所述。
图像输出单元，用于将实际视点图像和虚拟视点图像分别缩放至预定的尺度后生成对应的单视点图像并输出；具体如上方法实施例所 述。
所述的将二维图像转化为多视点图像的显示系统，其中，所述合成图像生成模块具体包括：
实际视点标注单元，用于依次存储单视点图像，并将单视点图像编号为1、2、...、N视点，并标注出实际视点，其中N为自然数；具体如上方法实施例所述。
像素插值单元，用于依次检查除实际视点外的其它单视点图像是否存在空洞区域，若存在空洞区域，则采用空洞区域周围的非空洞区域且与空洞区域相邻的四个像素、按照距离的远近按一定的比重进行像素插值和填充；具体如上方法实施例所述。
突变区域处理单元，用于依次检查除实际视点外的其它单视点图像是否有突变的区域，若存在突变的区域，提取突变区域的位置标记，并检查实际视点图像的相同位置标记对应的位置区域或临近区域是否存在同样的突变的区域，若实际视点图像存在同样的突变区域，则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理；具体如上方法实施例所述。
合成图像生成单元，用于根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅与裸眼3D显示屏的物理像素一一对应的合成图像；具体如上方法实施例所述。
所述的将二维图像转化为多视点图像的显示系统，其中，所述显示模块具体包括：
图像处理及显示单元，用于对合成图像进行帧频转换、Gamma校正、信号幅度调整、色域格式转换处理后，将处理后的合成图像转化成逻辑电信号，并发送至裸眼3D显示屏进行显示；具体如上方法实施例所述。
综上所述，本发明提供了一种将二维图像转化为多视点图像的显示方法及系统，方法包括：获取待处理的二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，根据深度值大小进行分层生成分层图像；通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，估算出虚拟视点图像的空白区域像素并填充空白区域，生成单视点图像输出；依次存储单视点图像，检测各个单视点图像的空洞区域检查并进行填充处理、检查突变区域后进行平滑处理，各个单视点图像集成为一幅合成图像；将合成图像处理后发送至裸眼3D显示屏显示。本发明能够实现二维图像转多视点图像、及进行裸眼3D显示，减少图像失真，简单方便、成本低。
应当理解的是，本发明的应用不限于上述的举例，对本领域普通技术人员来说，可以根据上述说明加以改进或变换，所有这些改进和变换都应属于本发明所附权利要求的保护范围。

Claims (10)

  1. 一种将二维图像转化为多视点图像的显示方法，其特征在于，方法包括；
    对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像；
    通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出；
    依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像；
    对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示。
  2. 根据权利要求1所述的将二维图像转化为多视点图像的显示方法，其特征在于，所述对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像具体包括：
    对待处理的二维图像进行锐化处理，搜索检测二维图像中的物体的边界和轮廓识别物体对象，分割二维图像中的物体对象并获取，将获取到的物体对象进行标识；
    获取各个物体对象的灰度分量，对灰度分量进行傅立叶变换后得到各个物体对象的频率分量，根据频率分量计算出各个物体对象的深度值；
    根据各个物体对象的深度值按照层与深度值对应、层间距为物体对象的深度值之差对物体对象进行分层，生成分层图像。
  3. 根据权利要求2所述的将二维图像转化为多视点图像的显示方法，其特征在于，所述通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出具体包括：
    通过预先设定的不同的视点位置观看分层图像，分别标定出视点图像中的实际视点位置和虚拟视点位置；
    获取标定的虚拟视点所在的物体对象的深度值，根据物体对象的深度值的差异采用不同的插值算法进行插值计算，根据计算后的结果获取标定的虚拟视点的空白区域的各个位置的像素后进行填充，去掉分层图像的层间距形成虚拟视点图像；
    将实际视点图像和虚拟视点图像分别缩放至预定的尺度后生成对应的单视点图像并输出。
  4. 根据权利要求3所述的将二维图像转化为多视点图像的显示方法，其特征在于，所述依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视 点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像具体包括：
    依次存储单视点图像，并将单视点图像编号为1、2、...、N视点，并标注出实际视点，其中N为自然数；
    依次检查除实际视点外的其它单视点图像是否存在空洞区域，若存在空洞区域，则采用空洞区域周围的非空洞区域且与空洞区域相邻的四个像素、按照距离的远近按一定的比重进行像素插值和填充；
    依次检查除实际视点外的其它单视点图像是否有突变的区域，若存在突变的区域，提取突变区域的位置标记，并检查实际视点图像的相同位置标记对应的位置区域或临近区域是否存在同样的突变的区域，若实际视点图像存在同样的突变区域，则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理；
    根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅与裸眼3D显示屏的物理像素一一对应的合成图像。
  5. 根据权利要求4所述的将二维图像转化为多视点图像的显示方法，其特征在于，所述对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示具体包括：
    对合成图像进行帧频转换、Gamma校正、信号幅度调整、色域格式转换处理后，将处理后的合成图像转化成逻辑电信号，并发送至裸眼3D显示屏进行显示。
  6. 一种将二维图像转化为多视点图像的显示系统，其特征在于， 系统包括：
    图像分层模块，用于对待处理的二维图像进行预处理后，获取二维图像中的物体对象并标识，根据物体对象的频率分量计算物体对象的深度值，按照深度值的大小进行分层生成分层图像；
    单视点图像输出模块，用于通过预先设定的不同视点位置观看分层图像，标定视点图像，根据分层图像的深度值差异，采用不同的插值算法估算出虚拟视点图像的空白区域像素并填充空白区域，去掉分层图像的层间距形成虚拟视点图像，对视点图像进行缩放，形成单视点图像输出；
    合成图像生成模块，用于依次存储单视点图像，将单视点图像进行编号，检测各个单视点图像的空洞区域并进行填充处理，并对各个单视点图像进行突变区域检查和平滑处理后，将各个单视点图像集成为一幅合成图像；
    显示模块，用于对合成图像进行输出处理后，将合成图像转化成逻辑电信号发送至裸眼3D显示屏进行显示。
  7. 根据权利要求6所述的将二维图像转化为多视点图像的显示系统，其特征在于，所述图像分层模块具体包括：
    标识单元，用于对待处理的二维图像进行锐化处理，搜索检测二维图像中的物体的边界和轮廓识别物体对象，分割二维图像中的物体对象并获取，将获取到的物体对象进行标识；
    深度值计算单元，用于获取各个物体对象的灰度分量，对灰度分量进行傅立叶变换后得到各个物体对象的频率分量，根据频率分量计 算出各个物体对象的深度值；
    分层单元，用于根据各个物体对象的深度值按照层与深度值对应、层间距为物体对象的深度值之差对物体对象进行分层，生成分层图像。
  8. 根据权利要求7所述的将二维图像转化为多视点图像的显示系统，其特征在于，所述单视点图像输出模块具体包括：
    视点标定单元，用于通过预先设定的不同的视点位置观看分层图像，分别标定出视点图像中的实际视点位置和虚拟视点位置；
    虚拟视点获取单元，用于获取标定的虚拟视点所在的物体对象的深度值，根据物体对象的深度值的差异采用不同的插值算法进行插值计算，根据计算后的结果获取标定的虚拟视点的空白区域的各个位置的像素后进行填充，去掉分层图像的层间距形成虚拟视点图像；
    图像输出单元，用于将实际视点图像和虚拟视点图像分别缩放至预定的尺度后生成对应的单视点图像并输出。
  9. 根据权利要求8所述的将二维图像转化为多视点图像的显示系统，其特征在于，所述合成图像生成模块具体包括：
    实际视点标注单元，用于依次存储单视点图像，并将单视点图像编号为1、2、...、N视点，并标注出实际视点，其中N为自然数；
    像素插值单元，用于依次检查除实际视点外的其它单视点图像是否存在空洞区域，若存在空洞区域，则采用空洞区域周围的非空洞区域且与空洞区域相邻的四个像素、按照距离的远近按一定的比重进行像素插值和填充；
    突变区域处理单元，用于依次检查除实际视点外的其它单视点图 像是否有突变的区域，若存在突变的区域，提取突变区域的位置标记，并检查实际视点图像的相同位置标记对应的位置区域或临近区域是否存在同样的突变的区域，若实际视点图像存在同样的突变区域，则检查的单视点图像突变区域是正常的，否则进行降噪平滑处理；
    合成图像生成单元，用于根据裸眼3D显示屏的物理像素排列情况，将N个单视点图像的像素进行交错排列，将N个单视点图像集成为一幅与裸眼3D显示屏的物理像素一一对应的合成图像。
  10. 根据权利要求9所述的将二维图像转化为多视点图像的显示系统，其特征在于，所述显示模块具体包括：
    图像处理及显示单元，用于对合成图像进行帧频转换、Gamma校正、信号幅度调整、色域格式转换处理后，将处理后的合成图像转化成逻辑电信号，并发送至裸眼3D显示屏进行显示。
PCT/CN2016/086466 2016-03-16 2016-06-20 一种将二维图像转化为多视点图像的显示方法及系统 Ceased WO2017156905A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
AU2016397878A AU2016397878B2 (en) 2016-03-16 2016-06-20 Display method and system for converting two-dimensional image into multi-viewpoint image
US15/759,714 US10334231B2 (en) 2016-03-16 2016-06-20 Display method and system for converting two-dimensional image into multi-viewpoint image

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610149174.3 2016-03-16
CN201610149174.3A CN105791803B (zh) 2016-03-16 2016-03-16 一种将二维图像转化为多视点图像的显示方法及系统

Publications (1)

Publication Number Publication Date
WO2017156905A1 true WO2017156905A1 (zh) 2017-09-21

Family

ID=56393754

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/086466 Ceased WO2017156905A1 (zh) 2016-03-16 2016-06-20 一种将二维图像转化为多视点图像的显示方法及系统

Country Status (4)

Country Link
US (1) US10334231B2 (zh)
CN (1) CN105791803B (zh)
AU (1) AU2016397878B2 (zh)
WO (1) WO2017156905A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110992304A (zh) * 2019-10-30 2020-04-10 浙江力邦合信智能制动系统股份有限公司 二维图像深度测量方法及其在车辆安全监测中的应用
CN115442580A (zh) * 2022-08-17 2022-12-06 深圳市纳晶云实业有限公司 一种便携式智能设备裸眼3d图片效果处理方法

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10230935B2 (en) * 2016-10-11 2019-03-12 Marvel Digital Limited Method and a system for generating depth information associated with an image
HK1224513A2 (zh) * 2016-10-14 2017-08-18 智能3D有限公司 通过机器学习技术改进2d至3d的自动转换质量的方法
CN106851247A (zh) * 2017-02-13 2017-06-13 浙江工商大学 基于深度信息的复杂场景分层方法
CN107155101A (zh) * 2017-06-20 2017-09-12 万维云视（上海）数码科技有限公司 一种3d播放器使用的3d视频的生成方法及装置
CN107580207A (zh) * 2017-10-31 2018-01-12 武汉华星光电技术有限公司 光场3d显示单元图像的生成方法及生成装置
CN108063894B (zh) * 2017-12-22 2020-05-12 维沃移动通信有限公司 一种视频处理方法及移动终端
CN111427166B (zh) * 2020-03-31 2022-07-05 京东方科技集团股份有限公司 一种光场显示方法及系统、存储介质和显示面板
CN112015357B (zh) * 2020-08-12 2023-05-05 浙江迅实科技有限公司 一种3d立体画的制作方法及其产品
KR20230145421A (ko) * 2021-04-04 2023-10-17 레이아 인코포레이티드 멀티뷰 이미지 생성 시스템 및 방법
CN113256544B (zh) * 2021-05-10 2023-09-26 中山大学 一种多视点图像合成方法、系统、装置及存储介质
CN120111200A (zh) * 2025-01-15 2025-06-06 北京天马辉电子技术有限责任公司 3d显示图像生成方法及相关设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101742349A (zh) * 2010-01-05 2010-06-16 浙江大学 一种对三维场景的表达方法及其电视系统
CN101902657A (zh) * 2010-07-16 2010-12-01 浙江大学 一种基于深度图分层的虚拟多视点图像的生成方法
US20130321576A1 (en) * 2012-06-01 2013-12-05 Alcatel-Lucent Methods and apparatus for encoding and decoding a multiview video stream
CN104837000A (zh) * 2015-04-17 2015-08-12 东南大学 一种利用轮廓感知的虚拟视点合成方法

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1641702A (zh) * 2004-01-13 2005-07-20 邓兴峰 由平面图象设计立体图象的方法
EP1889171A4 (en) * 2005-04-07 2012-11-28 Visionsense Ltd METHOD FOR RECONSTRUCTING A THREE-DIMENSIONAL SURFACE OF AN OBJECT
KR100957129B1 (ko) * 2008-06-12 2010-05-11 성영석 영상 변환 방법 및 장치
US20100259595A1 (en) * 2009-04-10 2010-10-14 Nokia Corporation Methods and Apparatuses for Efficient Streaming of Free View Point Video
WO2011104151A1 (en) * 2010-02-26 2011-09-01 Thomson Licensing Confidence map, method for generating the same and method for refining a disparity map
KR101071911B1 (ko) * 2010-10-29 2011-10-10 (주)포디비전 3차원 입체 영상 생성 방법
WO2012084277A1 (en) * 2010-12-22 2012-06-28 Thomson Licensing Apparatus and method for determining a disparity estimate
CN107346061B (zh) * 2012-08-21 2020-04-24 快图有限公司 用于使用阵列照相机捕捉的图像中的视差检测和校正的系统和方法
CN103269435A (zh) * 2013-04-19 2013-08-28 四川长虹电器股份有限公司 双目转多目虚拟视点合成方法
CN104820981B (zh) * 2015-04-22 2017-10-31 上海交通大学 一种基于视差分层分割的图像立体表示方法及系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101742349A (zh) * 2010-01-05 2010-06-16 浙江大学 一种对三维场景的表达方法及其电视系统
CN101902657A (zh) * 2010-07-16 2010-12-01 浙江大学 一种基于深度图分层的虚拟多视点图像的生成方法
US20130321576A1 (en) * 2012-06-01 2013-12-05 Alcatel-Lucent Methods and apparatus for encoding and decoding a multiview video stream
CN104837000A (zh) * 2015-04-17 2015-08-12 东南大学 一种利用轮廓感知的虚拟视点合成方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110992304A (zh) * 2019-10-30 2020-04-10 浙江力邦合信智能制动系统股份有限公司 二维图像深度测量方法及其在车辆安全监测中的应用
CN115442580A (zh) * 2022-08-17 2022-12-06 深圳市纳晶云实业有限公司 一种便携式智能设备裸眼3d图片效果处理方法
CN115442580B (zh) * 2022-08-17 2024-03-26 深圳市纳晶云实业有限公司 一种便携式智能设备裸眼3d图片效果处理方法

Also Published As

Publication number Publication date
CN105791803A (zh) 2016-07-20
AU2016397878A1 (en) 2018-04-19
US20190052864A1 (en) 2019-02-14
AU2016397878B2 (en) 2018-09-13
US10334231B2 (en) 2019-06-25
CN105791803B (zh) 2018-05-18

Similar Documents

Publication Publication Date Title
WO2017156905A1 (zh) 一种将二维图像转化为多视点图像的显示方法及系统
EP2992508B1 (en) Diminished and mediated reality effects from reconstruction
CN108335350A (zh) 双目立体视觉的三维重建方法
JP5442111B2 (ja) 画像から高速に立体構築を行なう方法
CN115176459B (zh) 虚拟视点合成方法、电子设备和计算机可读介质
CN101873509B (zh) 消除深度图序列背景和边缘抖动的方法
US9154762B2 (en) Stereoscopic image system utilizing pixel shifting and interpolation
CN103971408A (zh) 三维人脸模型生成系统及方法
KR100897542B1 (ko) 임의 시점 영상 합성 시 영상 보정 방법 및 장치
EP3633606B1 (en) Information processing device, information processing method, and program
CN103702103B (zh) 基于双目相机的光栅立体印刷图像合成方法
CN103559737A (zh) 一种对象全景建模方法
CN102665086A (zh) 利用基于区域的局部立体匹配获取视差的方法
CN115719320B (zh) 基于遥感影像的倾斜校正稠密匹配方法
CN106254854A (zh) 三维图像的获得方法、装置及系统
CN103369331B (zh) 图像空洞的填补方法和装置及视频图像的处理方法和装置
CN107147894B (zh) 一种自由立体显示中的虚拟视点图像生成方法
CN116051916A (zh) 训练数据获取方法、模型训练方法以及视差图像获取方法
CN109819231A (zh) 一种视觉自适应裸眼3d图像处理方法及装置
CN105657401B (zh) 一种裸眼3d显示方法、系统及裸眼3d显示装置
CN118898810A (zh) 输电线缆外破隐患识别方法、装置、电子设备和存储介质
CN117372647A (zh) 一种建筑用三维模型的快速构建方法及系统
KR101841750B1 (ko) 매칭 정보에 의한 입체 영상 보정 장치 및 그 방법
CN110068308A (zh) 一种基于多目相机的测距方法及测距系统
CN103945206A (zh) 一种基于相似帧比较的立体图像合成系统

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2016397878

Country of ref document: AU

Date of ref document: 20160620

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16894067

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 16894067

Country of ref document: EP

Kind code of ref document: A1