CN104091336B - Stereoscopic image synchronous segmentation method based on dense disparity map - Google Patents
Stereoscopic image synchronous segmentation method based on dense disparity map Download PDFInfo
- Publication number
- CN104091336B CN104091336B CN201410328103.0A CN201410328103A CN104091336B CN 104091336 B CN104091336 B CN 104091336B CN 201410328103 A CN201410328103 A CN 201410328103A CN 104091336 B CN104091336 B CN 104091336B
- Authority
- CN
- China
- Prior art keywords
- color
- image
- parallax
- background
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000011218 segmentation Effects 0.000 title claims abstract description 38
- 230000001360 synchronised effect Effects 0.000 title claims abstract description 10
- 238000009826 distribution Methods 0.000 claims abstract description 12
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 238000005457 optimization Methods 0.000 claims abstract description 4
- 239000003086 colorant Substances 0.000 claims description 15
- 230000002452 interceptive effect Effects 0.000 claims description 9
- 238000013179 statistical model Methods 0.000 claims description 9
- 238000005315 distribution function Methods 0.000 claims description 6
- 101150064138 MAP1 gene Proteins 0.000 claims description 2
- 238000010586 diagram Methods 0.000 claims description 2
- 101150077939 mapA gene Proteins 0.000 claims description 2
- 238000003709 image segmentation Methods 0.000 abstract description 6
- 230000000694 effects Effects 0.000 abstract description 5
- 230000001419 dependent effect Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000012733 comparative method Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a stereoscopic image synchronous segmentation method based on a dense disparity map. First, a group of stereoscopic images are input, and the disparity map is obtained through a stereoscopic image matching algorithm; then, part of the foreground and part of the background are appointed in the mode of sketching in one image through a brush; according to the appointed parts, prior statistic models of color distribution of the foreground and the background and prior statistic models of disparity distribution of the foreground and the background are established respectively; on the basis, constraints such as color, the gradient and the disparity are formalized in an image segmentation theoretical framework, and an energy function is constructed; finally, the optimization result is solved through the maximum flow/minimum segmentation algorithm of the image; if a user does not achieve the ideal effect, error areas in the image can keep being sketched till the ideal result is obtained. The disparity distribution models and the change models adopted for the method are all disparity statistic information, and thus influences caused by disparity calculation errors are effectively avoided. Compared with an existing method, the segmentation result obtained through the method is more accurate.
Description
Technical Field
The invention belongs to the cross field of computer vision, computer graphics, image processing and the like, and relates to a stereo image synchronous segmentation method based on a dense disparity map.
Background
The popularization of stereoscopic images in various fields puts an urgent need on intelligent processing of such data. Interactive stereoscopic image intelligent segmentation is one of the important tasks: a user only needs to appoint a small amount of front and background on one image in the stereo image, and the method can automatically complete the synchronous segmentation of the two images. The effect of the segmentation algorithm determines the accuracy of problems in video surveillance applications such as detection, identification, classification, and tracking. The segmented foreground target can be used as input data for three-dimensional model reconstruction, and the interference of the background in the reconstruction process is removed. The segmentation algorithm and program can help ordinary users to edit daily life pictures shot by the stereo camera and help movie producers to edit stereo televisions and movies in the later period. Such as removing unwanted objects, compositing foreground objects into a new background, and copying and pasting foreground objects, etc.
At present, interactive segmentation methods for single images are relatively mature, and some methods have already been put into practical use, for exampleThe Quick Selection (Quick Selection) tool in CS 3. Compared with the segmentation of a single image, the intelligent segmentation of the interactive stereoscopic image starts late. The existing basic framework for segmenting a stereo image is as follows: firstly, a disparity map is obtained through a stereo matching algorithm. Each pixel value in the disparity map represents an offset of a corresponding pixel in a reference map (one of two pre-selected maps) in a matching map. That is, given a pair of corresponding disparity maps of the stereo map and the left map, the corresponding pixel of the left image pixel in the right map can be obtained. After obtaining the disparity map, the energy function is formed by formalizing disparity clues and clues such as colors and gradients commonly used in single image segmentation. And solving the image segmentation problem by optimizing an energy function. The quality of the disparity map has an important influence on the segmentation result. However, the disparity map obtained by the conventional stereo matching method has many errors, and the conventional stereo image segmentation method based on the disparity map, for example, "StereoCut: in the dependent Interactive Object Selection in Stereo Image pages ", the correspondence determined by the disparity map is directly formalized in the energy function, which easily causes segmentation errors and affects the intellectualization of segmentation.
Disclosure of Invention
In view of the limitation of the current stereo image segmentation method based on the disparity map in the aspect of disparity use, the invention explores a new segmentation method under the theoretical framework of stereo image synchronous segmentation based on the disparity map, tries to reduce the influence of matching errors on segmentation results, and achieves the purpose of improving the intelligence of the segmentation process.
In order to realize the aim, the technical scheme of the invention is as follows: after a group of stereo images are input by a user, the method automatically obtains a disparity map through a stereo image matching algorithm. Then, the user can designate a part of the front and background in one of the figures by means of brush drawing. And respectively establishing prior statistical models of the color distribution of the front and background and prior statistical models of the parallax distribution of the front and background automatically according to the designated part. Based on the above, the constraints of color, gradient and parallax are formalized under the frame of graph-cut theory, and an energy function is constructed. And finally, solving an optimization result by adopting a maximum flow/minimum cut algorithm of the graph. If the user does not obtain the ideal effect, the user can also continue to draw the error area in the graph until an ideal result is obtained.
Compared with the prior art, the invention has the following advantages: the method takes the parallax map as a basis, establishes a front and background parallax distribution statistical model, mathematically formalizes the change condition of the parallax in the image, constructs an energy function by combining the traditional constraint terms, and solves the minimum value of the energy function through an image segmentation algorithm to realize segmentation. The parallax distribution model and the change model are parallax statistical information, and influence caused by parallax calculation errors is effectively avoided. Experiments prove that: compared with the prior art, the method provided by the invention has the advantage that the obtained segmentation result is more accurate on the premise of equal interaction amount.
Drawings
FIG. 1 is a flow chart of a method according to the present invention;
FIG. 2 shows experimental results of an application example of the present invention: (a) (b) are input left and right images, (c) and (d) are "StereoCut: the result of calculation by the method in the sensitive Interactive object selection in Stereo Image papers "; (e) and (f) is the segmentation result of the invention; the user inputs for both methods are shown in (c) and (e), with the solid lines inside the object identifying the foreground and the dashed lines outside the object area identifying the background.
Detailed Description
The invention is further described with reference to the following figures and detailed description.
The process of the invention is shown in fig. 1, and specifically comprises the following steps:
step one, matching a stereo image.
Reading in a pair of stereo images I ═ Il，Ir}，IlAnd IrRespectively representing left and right images. Calculating to obtain corresponding disparity maps of the left image and the right image by using a stereo matching algorithm, and respectively using DlAnd DrAnd (4) showing. Stereo matching may employ any algorithm, such as the algorithm proposed in the paper "efficiency Belief Propagation for EarlyVision" published by Felzenszwalb et al on CVPR 04.
And step two, adding front and background clues.
The user designates a part of the front and background in any one of the images through the designed interface. The embodiment of the invention adopts a method similar to that of 'StereoCut' published by Price et al on ICCV of 2011: the method used in the dependent InteractiveObject Selection in Stereo Image pages specifies the front and background pixels of a part by drawing lines of different colors on an Image using an input device such as a mouse, a touch panel, or a stylus. As shown in fig. 2(e), the pixels covered by the red lines belong to the foreground, and the pixels covered by the blue lines belong to the background. The subsequent steps of the present invention are not limited to the method of assigning the background and the foreground pixels used in the subsequent steps, and other methods can be used.
And step three, establishing color and parallax prior models of the front and the background.
The foreground pixel set designated by the user is represented by F, and the user is represented by BA specified set of background pixels. The prior models of the color and the parallax of the front and the background can be expressed in the form of GMM, histogram and a plurality of clusters, and are obtained by fitting or counting the color of the corresponding pixel set. The embodiment of the invention adopts a K-means clustering method to cluster the color values corresponding to the pixels in the F and the B respectively to obtain NcIndividual foreground color clusterMcIndividual background color clusterRespectively representing a statistical model of the color distribution of the background. Meanwhile, the parallax values corresponding to the pixels in the F and the B are respectively clustered by the same method to obtain NdIndividual foreground disparity clusterMdIndividual background parallax clusterAnd respectively representing the parallax distribution statistical models of the front and the background. The present invention proposes Nc＝Mc＝Nd＝Md＝64。
And step four, defining an energy function.
Stereo image I ═ { I ═ Il，IrIs composed of a left panel IlAnd right picture IrCan be expressed as an undirected graph G ═<ν,>. And v is a node set in the graph G and is a set of edges. Each vertex in the graph G corresponds to a pixel in the stereoscopic image I. The remaining pixels in the image, except for the pixels in set F and set B, are set to belong to set U. The synchronous segmentation of the interactive stereo image is carried out for each pixel p in U under the constraint of input strokesiAssigning a label xi。xi∈ {1,0} representing the front and background, respectively, the edges in graph G include the connecting edges of adjacent pixels in the image and the connections between corresponding points in the stereo image determined by the disparity mapAnd (7) edge.
Defining the solution of the stereo image synchronization problem as an optimization problem of the following objective energy function:
wherein f isUnary(pi，xi) Is a univariate Term (Unary Term) representing a pixel piThe similarity of the color and the parallax with the front and the background color and the parallax statistical model is also called as a Data Term (Data Term). The higher the similarity, fUnaryThe larger the value. f. ofIntra(pi，pj) Is an Intra-image Binary Term (Intra-Plane Binary Term) reflecting the differences between all pixels in I and their neighbourhoods (four neighbourhoods or eight neighbourhoods). N is a radical ofIntraAnd representing a set containing the adjacency relation of all the pixel points in the left and right graphs. The larger the difference, the smaller the term. According to the principle of graph cut algorithm, there is a tendency to take different labels between neighboring pixels at this time.The Term is a Binary Term (Inter Plane Binary Term) between images, and defines the matching result of the corresponding point, and the Term is larger as the matching degree is higher. CInterA set containing left and right pixel point Correspondence (coresponsondence) relationships is shown. Lambda [ alpha ]Unary，λIntra，λInterIs to adjust the weights between the energy terms.
(1) Defining unary constraint terms
The univariate constraint item comprises a color univariate item and a parallax univariate item, and is defined as follows:
fUnary(pi，xi)＝λc(1-Pc(xi|ci))+λd(1-Pd(x|di)) (2)
wherein, Pc(xi|ci) To representGiven pixel piColor c ofi，xiProbability values for foreground or background labels are taken. Considering that the larger the probability, the smaller the energy function should be, so take 1-PcRepresenting a color unary item. Likewise, Pc(xi|di) Representing a given pixel piThe parallax value d ofi，xiProbability values for foreground or background labels are taken. Taking 1-PdRepresenting a parallax entry. Lambda [ alpha ]c、λdRespectively representing the influence weights, lambda, of color and parallaxc+λd＝1。
The invention represents the color and parallax models of the front and background in cluster-like form (N)cIndividual foreground color clusterMcIndividual background color clusterNdIndividual foreground disparity clusterMdIndividual background parallax cluster) And a calculation method of the unary item is given.
The color unary is calculated as follows. And comparing the color of each unmarked pixel with the cluster classes of the foreground and background colors, and finding the minimum distance between the color of each unmarked pixel and the center of the cluster classes, wherein the distance is used for describing the similarity of the pixel color with the front and background colors. The smaller the distance from the foreground (or background) color, the closer the color, and the more likely the pixel is to select a foreground (or background) label according to graph cut theory. The mathematical form of the color unary term is described as:
wherein,respectively representing a pixel piColor c ofiThe minimum distances to the centers of various clusters of foreground and background colors are respectively expressed as:
the parallax unary item is the same as the color unary item in the calculation process.
(2) Defining intra-image binary constraint terms
Intra-image binary constraint term fIntra(pi，pj) The method comprises two terms, which respectively describe color change (i.e. color gradient) and parallax change (i.e. parallax gradient) around a pixel point, and is defined as follows:
fIntra(pi，pj)＝fc(pi，pj)fd(pi，pj) (4)
wherein f isc(pi，pj) Representing the similarity of colors between adjacent pixels, the closer the color is, the higher the value is, and the less the probability that the boundary will pass through the two is according to the principle of graph cut algorithm. f. ofd(pi，pj) Representing a pixel piRelative to adjacent pixel point pjThe similarity of the parallaxes. The closer the parallax is, the larger the value is, and the less the probability that the boundary will pass through the two is according to the principle of the graph cut algorithm. The invention proposes the following for two defined forms:
the two items can also take other forms, such as "StereoCut: exponential form used in the relationship Interactive Object Selection in Stereo Image pages ".
In fact, the disparity calculation has errors, and direct use introduces errors into the segmentation process. The proposed solution is to replace the disparity variation between two pixels with the disparity variation of the local area. Order SjRepresents piThe area of the site. Region SjThe variance var (S) is adopted for the inner parallax changej) And (4) showing. In this case, equation (6) becomes:
wherein S isj＝A(pi) Function A (p)i) Representing a pixel piThe area in which it is located. The image area may be obtained by an over-segmentation method, or by segmenting the picture into a small set of square areas in advance.
(3) Defining inter-image binary constraint terms
Corresponding pixels between the two elements constrained stereo images in the images take the same label, and are defined as follows:
wherein C represents in a stereoscopic imageAs corresponding points betweenPerformance, is an asymmetric function:
is determined based on a disparity mapAs a function of the probability distribution of the corresponding points. Function(s)To representIs a left image pixelAnd determining the corresponding relation according to the disparity map at the corresponding point on the right image. Price et al published on ICCV of 2011 as "StereoCut: the relationship Interactive Object Selection in Stereo Image PairsThe five defining modes of (1) are uniform distribution, Delta function, probability density distribution function, consistent Delta function and consistent probability density distribution function. Because of the high computational complexity of the forms of uniform distribution, probability density distribution function, consistent probability density distribution function, the present invention proposes to use a Delta function or consistent Delta function, defined as follows. The Delta function uses only the disparity map corresponding to a single map. Let { dlThe left image disparity set is defined as follows:
wherein,is a pixel in the left imageCorresponding points in the right pictureThe parallax error of (1). Let { drThe right view disparity set is given by the consistent Delta function:
wherein,is a pixel in the right picturePoints corresponding to the left graphThe parallax error of (1).
In the formula (9)To representAndthe color similarity between them, in case the parallax is completely accurate,however, the current parallax calculation method has errors, and the proposed form of the invention is as follows:
wherein,is a left image pixelThe color value of (a) of (b),is thatCorresponding point of the right diagramThe color value of (a).
And step five, solving the minimum value of the energy function.
The present invention employs a graph Cut algorithm, such as the Max Flow/Min Cut algorithm proposed in the paper "An Experimental company of Min-Cut/Max-Flow Algorithms for Energy Minimization in Vision" published by Yuri Boykov et al in IEEE Transaction on PAMI, 2004, to optimize the Energy function (equation (1)) defined by the present invention to obtain the optimal labeling result, i.e., the segmentation result. If the user is not satisfied with the segmentation result, the step two can be returned to, and the pre-background clues and the background clues are continuously added. Each additional stroke triggers a complete segmentation process.
"StereoCut: the method in the dependent InteractiveObject Selection in the Stereo Image papers "is a comparison object, and the effectiveness of the method of the present invention will be described. Both methods use a consistent Delta function (equation (11)) as the probability distribution function between corresponding points. Figure 2 shows a comparison of the effects. Fig. 2(a) and (b) show the input left and right images. (c) And (d) the result calculated by adopting a StereoCut method; FIGS. 2(e), (f) are the segmentation results of the present invention; in the four result graphs, the closed curve around the object (street lamp) marks the boundary of the segmentation object; the user inputs for both methods are shown in figures (c), (e) respectively, with the solid lines inside the object marking the foreground and the dashed lines outside the object marking the background. Comparing the graphs (c) and (d) with the graphs (e) and (f), the method of the invention can obtain better segmentation effect under the premise of equal interaction amount. The contrast method requires the user to specify more foreground and background if a better result is desired. Therefore, the method of the present invention is more intelligent than the comparative method.
Claims (2)
1. A stereo image synchronous segmentation method based on a dense disparity map is characterized by comprising the following steps:
step one, matching a stereo image;
reading in a pair of stereo images I ═ Il，Ir}，IlAnd IrRespectively representing a left image and a right image; calculating to obtain corresponding disparity maps of the left image and the right image by using a stereo matching algorithm, and respectively using DlAnd DrRepresents;
adding front and background clues;
appointing a part of front and background in any one of the images through a designed interface; using input equipment such as a mouse, a touch screen or a handwriting pen and the like to designate partial front and background pixels by drawing lines with different colors on the image;
establishing prior models of the color and the parallax of the front and the background;
f represents a foreground pixel set specified by a user, and B represents a background pixel set specified by the user; the prior models of the color and the parallax of the front and the background are expressed in a form of a plurality of clusters, and are obtained by fitting or counting the colors of corresponding pixel sets: clustering color values corresponding to pixels in F and B respectively by utilizing a K-means algorithm to obtain NcIndividual foreground color clusterMcIndividual background color clusterA color distribution statistical model respectively representing a background; meanwhile, the parallax values corresponding to the pixels in the F and the B are respectively clustered by the same method to obtain NdIndividual foreground disparity clusterMdIndividual background parallax clusterRespectively representing the parallax distribution statistical models of the front and the background; n is a radical ofc＝Mc＝Nd＝Md＝64；
Step four, defining an energy function;
stereo image I ═ { I ═ Il,IrIs composed of a left panel IlAnd right picture IrCan be expressed as an undirected graph G ═<ν,>(ii) a V is a node set in the graph G and is a set of edges; each vertex in graph G corresponds to a pixel in stereoscopic image I; in the image except for pixels in set F and set BThe rest pixels are set to belong to a set U; the synchronous segmentation of the interactive stereo image is carried out for each pixel p in U under the constraint of input strokesiAssigning a label xi；xi∈ {1,0} representing the front and background, respectively, the edges in graph G include the connecting edges of adjacent pixels in the image and the connecting edges between corresponding points in the stereo image determined by the disparity map;
defining the solution of the stereo image synchronization problem as an optimization problem of the following objective energy function:
wherein f isUnary(pi,xi) Is a univariate term representing a pixel piThe similarity of the color and the parallax with the front and background colors and the parallax statistical model is also called as a data item; the higher the similarity, fUnaryThe larger the value; f. ofIntra(pi,pj) The image is an intra-image binary item, the difference between all pixels in the image I and four adjacent domains or eight adjacent domains is reflected, and the larger the difference is, the smaller the item is; n is a radical ofIntraRepresenting a set containing the adjacent relation of all pixel points in the left and right images; according to the principle of graph cut algorithm, different labels tend to be taken among the neighborhood pixels at the moment;the binary item between the images defines the matching result of the corresponding point, and the higher the matching degree is, the larger the item is; cInterRepresenting a set containing left and right image pixel point correspondences; lambda [ alpha ]Unary,λIntra,λInterAdjusting the weight among the energy items;
(1) defining unary constraint terms
The univariate constraint item comprises a color univariate item and a parallax univariate item, and is defined as follows:
fUnary(pi,xi)＝λc(1-Pc(xi|ci))+λd(1-Pd(x|di)) (2)
wherein, Pc(xi|ci) Representing a given pixel piColor c ofi，xiTaking the probability value of the foreground or background label; considering that the larger the probability, the smaller the energy function should be, so take 1-PcRepresenting a color unary; likewise, Pc(xi|di) Representing a given pixel piThe parallax value d ofi，xiTaking the probability value of the foreground or background label; taking 1-PdRepresenting a disparity unary; lambda [ alpha ]c、λdRespectively representing the influence weights, lambda, of color and parallaxc+λd＝1；
Representing the color and parallax models of the front and background in cluster-like form, including NcIndividual foreground color clusterMcIndividual background color clusterNdIndividual foreground disparity clusterMdIndividual background parallax clusterGiving a calculation method of the unary item;
the color unary is calculated as follows: comparing the color of each unmarked pixel with the cluster of the foreground and background colors, and finding the minimum distance between the color of each unmarked pixel and the center of the cluster, wherein the distance is used for describing the similarity between the pixel color and the front and background colors; the smaller the distance from the foreground or background color is, the closer the color is, and according to the graph cut theory, the more the pixel tends to select the foreground or background label; the mathematical form of the color unary term is described as:
wherein,respectively representing a pixel piColor c ofiThe minimum distances to the centers of various clusters of foreground and background colors are respectively expressed as:
the parallax unary item and the color unary item are calculated in the same process;
(2) defining intra-image binary constraint terms
Intra-image binary constraint term fIntra(pi,pj) The method comprises two terms, which are used for respectively describing color change and parallax change around a pixel point, namely color gradient and parallax gradient, and are defined as follows:
fIntra(pi,pj)＝fc(pi,pj)fd(pi,pj) (4)
wherein f isc(pi,pj) Representing the similarity of colors between adjacent pixels, wherein the closer the colors are, the larger the value of the colors is, and the probability that the boundary passes through the two pixels is lower according to the principle of a graph cut algorithm; f. ofd(pi,pj) Representing a pixel piRelative to adjacent pixel point pjThe similarity of the parallaxes; the closer the parallax difference between the two is, the larger the value of the parallax difference is, and the probability that the boundary passes through the two is lower according to the principle of the graph cut algorithm; the two terms are defined as follows:
in fact, parallax calculation has errors, and direct use introduces errors into the segmentation process; the solution is to replace the parallax change between every two pixels with the parallax change of a local area; order SjRepresents piThe area where the device is located; region SjThe variance var (S) is adopted for the inner parallax changej) Represents; in this case, equation (6) becomes:
wherein S isj＝A(pi) Function A (p)i) Representing a pixel piThe area in which the device is located; the image area can be obtained by adopting an over-segmentation method, or a picture can be segmented into a small square area set in advance;
(3) defining inter-image binary constraint terms
Corresponding pixels between the two elements constrained stereo images in the images take the same label, and are defined as follows:
wherein C represents in a stereoscopic imageThe probability of being a corresponding point between is an asymmetric function:
is determined based on a disparity mapAs a probability distribution function of corresponding points; function(s)To representIs a left image pixelCorresponding points on the right image are determined according to the disparity map;a Delta function or a consistent Delta function is adopted, and the definition mode is as follows; the Delta function only uses a disparity map corresponding to a single map; let { dlThe left image disparity set is defined as follows:
wherein,is a pixel in the left imageCorresponding points in the right pictureThe parallax of (1); let { drThe right view disparity set is given by the consistent Delta function:
wherein,is a pixel in the right picturePoints corresponding to the left graphThe parallax of (1);
in the formula (9)To representAndthe color similarity between them, in case the parallax is completely accurate,however, the current parallax calculation method has errors, and the following forms are adopted for eliminating the errors:
wherein,is a left image pixelThe color value of (a) of (b),is thatCorresponding point of the right diagramA color value of (a);
step five, solving the minimum value of the energy function;
obtaining an optimal marking result, namely a segmentation result, by optimizing an energy function defined by the invention, namely an equation (1), by adopting a graph cutting algorithm; if the segmentation result is not satisfied, returning to the step two, and continuing to add the front and background clues; each additional stroke triggers a complete segmentation process.
2. The method for synchronous segmentation of stereoscopic images based on dense disparity maps according to claim 1, wherein the color and disparity prior models of the foreground and the background in step three can be further expressed in the form of GMM and histogram.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410328103.0A CN104091336B (en) | 2014-07-10 | 2014-07-10 | Stereoscopic image synchronous segmentation method based on dense disparity map |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410328103.0A CN104091336B (en) | 2014-07-10 | 2014-07-10 | Stereoscopic image synchronous segmentation method based on dense disparity map |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104091336A CN104091336A (en) | 2014-10-08 |
| CN104091336B true CN104091336B (en) | 2017-05-17 |
Family
ID=51639051
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410328103.0A Active CN104091336B (en) | 2014-07-10 | 2014-07-10 | Stereoscopic image synchronous segmentation method based on dense disparity map |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104091336B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104835146A (en) * | 2015-04-14 | 2015-08-12 | 上海大学 | Salient object segmenting method in stereo image based on depth information and image cutting |
| CN105046689B (en) * | 2015-06-24 | 2017-12-15 | 北京工业大学 | A kind of interactive stereo-picture fast partition method based on multi-level graph structure |
| CN105894519A (en) * | 2016-04-25 | 2016-08-24 | 武汉工程大学 | Robustness image segmentation algorithm based on low rank recovery |
| CN106650744B (en) * | 2016-09-16 | 2019-08-09 | 北京航空航天大学 | The image object of local shape migration guidance is divided into segmentation method |
| CN107492101B (en) * | 2017-09-07 | 2020-06-05 | 四川大学 | Multi-modal nasopharyngeal tumor segmentation algorithm based on self-adaptive constructed optimal graph |
| CN108230338B (en) * | 2018-01-11 | 2021-09-28 | 温州大学 | Stereo image segmentation method based on convolutional neural network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7991228B2 (en) * | 2005-08-02 | 2011-08-02 | Microsoft Corporation | Stereo image segmentation |
| CN103606162A (en) * | 2013-12-04 | 2014-02-26 | 福州大学 | Stereo matching algorithm based on image segmentation |
-
2014
- 2014-07-10 CN CN201410328103.0A patent/CN104091336B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7991228B2 (en) * | 2005-08-02 | 2011-08-02 | Microsoft Corporation | Stereo image segmentation |
| CN103606162A (en) * | 2013-12-04 | 2014-02-26 | 福州大学 | Stereo matching algorithm based on image segmentation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104091336A (en) | 2014-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104091336B (en) | Stereoscopic image synchronous segmentation method based on dense disparity map | |
| Wang et al. | Saliency-aware geodesic video object segmentation | |
| CN110111338B (en) | Visual tracking method based on superpixel space-time saliency segmentation | |
| Kennedy et al. | Optical flow with geometric occlusion estimation and fusion of multiple frames | |
| Vazquez-Reina et al. | Multiple hypothesis video segmentation from superpixel flows | |
| CN108537239B (en) | Method for detecting image saliency target | |
| WO2020206850A1 (en) | Image annotation method and device employing high-dimensional image | |
| Zhu et al. | Targeting accurate object extraction from an image: A comprehensive study of natural image matting | |
| Xu et al. | Automatic building rooftop extraction from aerial images via hierarchical RGB-D priors | |
| US9626585B2 (en) | Composition modeling for photo retrieval through geometric image segmentation | |
| WO2017181892A1 (en) | Foreground segmentation method and device | |
| CN104166988B (en) | A kind of stereo sync dividing method for incorporating sparse match information | |
| Li et al. | Motion-aware knn laplacian for video matting | |
| CN105046689B (en) | A kind of interactive stereo-picture fast partition method based on multi-level graph structure | |
| Yang et al. | A bundled-optimization model of multiview dense depth map synthesis for dynamic scene reconstruction | |
| Hu et al. | Markov random fields for sketch based video retrieval | |
| CN110163873B (en) | Bilateral video target segmentation method and system | |
| Pan et al. | Automatic segmentation of point clouds from multi-view reconstruction using graph-cut | |
| Zhao et al. | Real-time and temporal-coherent foreground extraction with commodity RGBD camera | |
| Tian et al. | HPM-TDP: An efficient hierarchical PatchMatch depth estimation approach using tree dynamic programming | |
| Mukherjee et al. | A hybrid algorithm for disparity calculation from sparse disparity estimates based on stereo vision | |
| Qi et al. | High-speed video salient object detection with temporal propagation using correlation filter | |
| Nguyen et al. | Interactive object segmentation from multi-view images | |
| Mirkamali et al. | RGBD image segmentation | |
| Wang et al. | Image segmentation incorporating double-mask via graph cuts |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |