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
  • H04N13/398—Synchronisation thereof; Control thereof
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
  • G06T7/50—Depth or shape recovery
  • G06T7/55—Depth or shape recovery from multiple images
  • G06T7/593—Depth or shape recovery from multiple images from stereo images
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
  • G06T7/97—Determining parameters from multiple pictures
• • 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
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/10—Image acquisition modality
  • G06T2207/10004—Still image; Photographic image
  • G06T2207/10012—Stereo images
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/20—Special algorithmic details
  • G06T2207/20068—Projection on vertical or horizontal image axis

Definitions

• the present invention relates to the field of scopic stereo vision systems and more particularly to the management of the position of the focal plane in a scopic stereo scene.
• Stereoscopic images are typically composed of a left image and a right image of the same scene. These two images are taken by cameras from a slightly shifted point of view. Generally, the two cameras are parallel and spaced a few centimeters apart.
• a scopic stereo video is composed of two video streams, a left video stream and a right video stream.
• Stereoscopy is due to the parallax effect. This effect results in the same object appearing in the two images, left and right, slightly translated. The value of the translation depends on the depth, that is, the distance to the cameras, from the object.
• an object is viewed on a stereo scopic screen 1.3.
• the eyes 1.1 and 1.2 of the user are represented.
• the object is located on the left image at point 1.6 and on the right image at point 1.5. These points are distant from a translation value 1.7, this value expressed in translation in the right image relative to the left image is a negative translation.
• this value expressed in translation in the right image relative to the left image is a negative translation.
• an object is viewed on a stereoscopic screen 2.3.
• the eyes 2.1 and 2.2 of the user are represented.
• the object is located on the left image at point 2.6 and on the right image at point 2.5. These points are distant from a translation value 2.7, this value expressed in translation in the right image relative to the left image is a positive translation.
• this point When the eyes of the user will accommodate to visualize this point, it will appear in point 2.4 as being behind the screen.
• an object is viewed on a stereoscopic screen 3.3.
• the eyes 3. 1 and 3.2 of the user are represented.
• the object is on the left image at point 3.6 and on the right image at point 3.5. These points merge, the translation for this point is zero between the two images.
• the eyes of the user will accommodate to visualize this point, it will appear in point 3.4 as being at the level of the screen.
• the focal plane of a stereoscopic scene is the plane perpendicular to the optical axes of the cameras located at a given depth such that all the points representing objects in this plane are represented in the left and right images with zero parallax.
• the optical axes of the cameras performing the stereoscopic shooting are typically parallel, the objects at infinity are superimposed perfectly in the two images while the closer objects undergo a translation all the stronger as the object is close to the cameras. .
• the images taken by such a system therefore have a focal plane to infinity.
• the entire scene is located in front of this screen. It is found that the visualization of a stereoscopic video where the focal plane is thus located at infinity tends to be uncomfortable to the eye.
• a focal plane located globally in front of the stage with rare objects that can be located in front of the focal plane is much more comfortable to visualize.
• the aim of the invention is to propose a method for managing the focal plane within an image or sequence of stereo scopic images.
• a translation of the right and left images is performed to place the focal plane in front of the visualized scene.
• the depth of the major objects of the scene is estimated by dividing the images into a plurality of zones and estimating the depth of each zone. The depth of the focal plane is then calculated so that most of the scene is behind the focal plane.
• the invention relates to a method for managing the position of the focal plane in a stereoscopic scene represented by a left image and a right image which comprises a step of determining at least one zone in the two images; a step of calculating the dominant translation value for each of the zones; a step of determining a relative translation value between the left and right images according to the dominant translation values of each of the zones and a relative translation step of the two left and right images of said previously determined value.
• the step of calculating the dominant translation value for each of the zones comprises a step of projecting the left zone and the right zone on a left line and a straight line; a step of calculating the correlation between the left line and the straight line for a plurality of translation values and a step of determining the translation value for which the correlation is the best.
• the step of determining a relative translation value between the left and right images comprises a step of classifying the dominant translation values of each zone by group of close values; a step of calculating the average value of the dominant translation values within each group and a step of choosing the highest average value that corresponds to the nearest group, as the relative translation value between the left and right images.
• the method further comprises a step of eliminating areas for which the dominant translation value obtained is aberrant.
• the method further comprises a step of eliminating the groups of zones not containing a number of zones at least equal to a given threshold.
• the invention also relates to a device for shooting or stereoscopic viewing, characterized in that it comprises means for managing the position of the focal plane in a stereoscopic scene represented by a left image and a right image comprising means for determining at least one area in both images; means for calculating the dominant translation value for each of the zones; means for determining a relative translation value between the left and right images based on the dominant translation values of each of the zones and means for performing a relative translation of the two left and right images of said previously determined value.
• Fig. 1 illustrates the positions of a point appearing behind the focal plane.
• Fig. 2 illustrates the positions of a point appearing in front of the focal plane.
• Fig. 3 illustrates the positions of a point appearing in the focal plane.
• Fig. 4 illustrates the method of displacement of the focal plane in the embodiment of the invention.
• Fig. 5 illustrates an exemplary embodiment of the focal plane management method.
• Fig. 6 illustrates a zone division of an image.
• Fig. 7 illustrates the projection on a line of a zone.
• FIG. 4 illustrates the method of displacement of the focal plane.
• a scene includes objects 4.1, 4.2 and 4.3 each located at a given depth as viewed by an observer 4.4.
• the scene is displayed on a 4.5 screen.
• the focal plane of the scene merges with the screen.
• Objects appear in front of the screen or behind the screen if they are in front of the focal plane or behind the focal plane.
• the scene is created from a left image 4.7 and a right image 4.8 which each represent the scene and its objects.
• the focal plane corresponds to the objes whose location in the left and right images is superimposed. Objects out of the focal plane undergo a parallax, negative for objects in front of the focal plane and positive for objects behind the focal plane.
• the cameras When shooting, the cameras have a parallel optical axis. As a result, the left and right images are superimposed for objects at infinity. When such scenes are viewed on the screen, the entire scene appears in front of the focal plane and thus as "out" of the screen. The visualization then requires constant adjustments of accommodation in an area close to the eyes and is uncomfortable.
• the relative translation value that must be applied to the left and right images to move the focal plane and position it to a given depth value is calculated based on a measurement of the depths of the objects present in the scene. When these depths have been measured, a depth is determined so that most of the scene is behind the focal plane. This is considered to be the case, for example, when more than 50% of the surface of the images appears as located behind the focal plane.
• the depth is not necessary to actually calculate the depth values. Since the depth is related to the parallax value and therefore to the translation value, it is sufficient to calculate the translation values for the different objects of the scene and to apply a translation value relative to the images calculated as a function of these values of translation for objects.
• Several focal plane management policies can be adopted here. According to a first exemplary embodiment, it is sought to place the focal plane so that most of the scene is behind the focal plane.
• FIG. 5 An exemplary embodiment of the focal plane management method according to this first embodiment is illustrated in FIG. 5.
• a first step 5.1 at least one zone is determined in the left and right images. Indeed, according to some embodiments, the entire image is analyzed or focuses on an area of interest in the image, typically a central area. According to other embodiments, the images are divided into a plurality of zones. The values, called Z, of dominant translation of each of the zones of the image are then calculated. The cutting can be done, for example, according to the method illustrated in FIG. 6. In the exemplary embodiment, an image 6.1 is divided into a set of zones 6.2 here determined by a 9-zone fair grid.
• Each zone is projected on a line according to the diagram of FIG. 7.
• An area 7.2 or each column is summed according to the operation 7.3 to give a single line 7.4. In this way we obtain a line that is representative of the area along the horizontal axis.
• the correlation can be done by any other means, for example by sum of the squares of the differences.
• the translation can also be determined by calculating the Fourier transform of the left and right lines, which is then divided into members. The inverse Fourier transform is then calculated. The index of the maximum standard value in the sense of the complex numbers then gives the optimal translation value.
• the translation can also be determined by direct calculation of parallaxes in two dimensions, for example by comparison of sub-blocks of the image, ie without prior projection of the images on lines.
• This method is naturally more precise but slower. Correlation on the projection of each zone on a line brings a gain in terms of computation time compared to a correlation made over the entire zone. Since the translation we are trying to measure is only a horizontal translation, the result is reliable. This is particularly important when the process is applied in real time when rendering the scene on a device with a limited computing resource.
• a step 5.2 it is advantageous to eliminate the areas for which the dominant translation value obtained is aberrant if there is one. That is, values that are too small or too large or too different from the other values obtained. This step is optional.
• the translation values obtained are grouped into groups of similar values.
• the idea here is to group areas of the image representing close objects in terms of depth.
• a threshold is fixed on the minimum size of a group in number of zones during a step 5.4. Only those groups with a minimum number of members are retained. For example, groups consisting of a single element are eliminated. We try to divide the image into significant groups of depths. We focus on objects of a certain size on the screen.
• step 5.5 the averages of the different dominant translation values obtained for each zone are calculated. We then choose the highest average value that corresponds to the closest significant group. In this way, the focal plane is placed in front of the stage. It is always possible that certain objects, for example eliminated during the thresholding operation 5.4, or too small to significantly influence the translation of a given zone, are in front of the focal plane. However, most of the scene is thus positioned behind the focal plane.
• step 5.6 the relative translation between the left and right images is then performed to effectively place the focal plane at the determined depth.
• This translation can be performed on the right image relative to the left image, or on the left image relative to the right image. It is also possible to distribute the translation on both images.
• it is sought to place the main object of the scene at the focal plane. To do this, for example, we define a single zone in the image, typically a central zone. The depth of this zone, or simply the corresponding translation, is then calculated as before and it is applied to the images to place the focal plane at this zone. Alternatively, one can also choose to place the focal plane at the depth level of the group with the most members in the algorithm described above.
• This method can be used on a still image consisting of a single left image and a single right image. It can also be used on a stereoscopic image sequence. In the latter case, it can be performed as processing on the sequence outside the viewing process or be performed in real time during viewing or when shooting.
• This last embodiment is particularly interesting when it is integrated in a camera for stereoscopic shooting and rendering. It can, for example, integrate in a device such as a mobile phone incorporating a stereoscopic view system, a portable game console, or a stereoscopic camera.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Theoretical Computer Science (AREA)
• Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
• Processing Or Creating Images (AREA)
• Image Analysis (AREA)

Priority Applications (3)

Applications Claiming Priority (2)

Publications (1)

Family

ID=44508464

Family Applications (1)

Country Status (5)

Families Citing this family (1)

Citations (5)

Family Cites Families (9)

• 2011
  • 2011-02-23 FR FR1151467A patent/FR2971875B1/fr active Active
• 2012
  • 2012-02-21 WO PCT/EP2012/052943 patent/WO2012113797A1/fr not_active Ceased
  • 2012-02-21 US US14/000,462 patent/US9800865B2/en active Active
  • 2012-02-21 JP JP2013554871A patent/JP5981460B2/ja active Active
  • 2012-02-21 EP EP12704435.2A patent/EP2678823B1/fr active Active

Patent Citations (5)

Also Published As

Similar Documents

Legal Events