CN103679686A - Match measure determination method, match measure determination device, parallax computation method and image matching method - Google Patents
Match measure determination method, match measure determination device, parallax computation method and image matching method Download PDFInfo
- Publication number
- CN103679686A CN103679686A CN201210335757.7A CN201210335757A CN103679686A CN 103679686 A CN103679686 A CN 103679686A CN 201210335757 A CN201210335757 A CN 201210335757A CN 103679686 A CN103679686 A CN 103679686A
- Authority
- CN
- China
- Prior art keywords
- image
- image block
- matching
- reference picture
- significance
- 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.)
- Granted
Links
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a match measure determination method, a match measure determination device, an image matching method and a parallax computation method which are used for stereoscopic vision matching. The match measure determination method can include the following steps: obtaining an initial match measure between a reference image block of a reference image and an object image block of a matching image; determining the significance of the reference image block of the reference image, which represents the significance degree of the reference image block among all image blocks in a whole or partial area of the reference image; adjusting the obtained initial match measure between the reference image block of the reference image and the object image block of the matching image based on the significance of the reference image block of the reference image, so as to determine the match measure between the reference image block of the reference image and the object image block of the matching image. Through the invention, the match measure which is more applicable for stereoscopic vision matching, more objective and more consistent with the actual condition can be obtained.
Description
Technical field
The present invention relates generally to image and processes, and relates more specifically to towards match measure method and apparatus, parallax calculation method, the image matching method of stereoscopic vision coupling.
Background technology
Images match is one of the basic research problems in computer vision and image understanding field, has been widely used in the numerous areas such as multi-source image fusion, target identification, three-dimensional reconstruction.
In images match process, relate to determining of match measure.In brief, match measure can be understood as the similar or difference degree of two pixels or image block, and this area similar term of another implication is to measure cost (cost), and wherein two pixels or image block are more dissimilar, and corresponding measurement cost is higher.
Classic method comprises the absolute difference and (Sum of Absolute Differences), the difference of two squares and (Sum of Squared Differences), normalization relevant (Normalized Cross-Correlation) etc. of often using.
In addition, be entitled as " Depth Discontinuities by Pixel-to-Pixel Stereo " at Birchfield and Tomasi, the IEEE International Conference on Computer Vision, Bombay, India, in 1998 article, propose to utilize linear interpolation to reduce to estimate the susceptibility to image sampling effect.
The open US No.7333651B1 of United States Patent (USP) has proposed a kind of match measure method based on stationary window.The method is passed through the quantity of matched pixel in the involved stationary window of the corresponding pixel of statistics, the degree of measuring pixel matching.
Summary of the invention
According to embodiments of the invention, a kind of reference image block of definite reference picture towards stereoscopic vision coupling and the method for the match measure between the object images piece of matching image are provided, wherein reference picture and matching image are divided into respectively image block, and this match measure determines that method can comprise: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
According to another embodiment of the present invention, a kind of reference image block of definite reference picture towards stereoscopic vision coupling and the device of the match measure between the object images piece of matching image are provided, wherein reference picture and matching image are divided into respectively image block, this match measure determining device can comprise: initial matching is estimated acquisition parts, and the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Reference image block significance determining means, determines the significance of this reference image block of this reference picture, and the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; Match measure adjustment component, the significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
According to an embodiment more of the present invention, a kind of image matching method for binocular stereo vision is provided, can comprise: (1) inputs left image and right image; (2) by left image and right image over-segmentation, be image block respectively; (3) one of left image and right image are used as to reference picture, another is as matching image, object images piece for the reference image block in reference picture and matching image, determine the match measure of the two: (4) match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates.Wherein can according to the match measure computing method of the embodiment of the present invention, calculate the match measure of two image blocks by what describe above.Particularly, can determine the match measure between the reference image block of reference picture and the object images piece of matching image by following operation: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
According to another embodiment of the invention, provide the parallax calculation method for binocular stereo vision according to the embodiment of the present invention, can comprise the steps: that (1) input left image and right image; (2) by left image and right image over-segmentation, be image block respectively; (3) one of left image and right image are used as to reference picture, another is as matching image, object images piece for the reference image block in reference picture and matching image, determine the match measure of the two: (4) match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates; (5) reference image block based in reference picture and the matching image piece in matching image, the parallax that computing reference image block is corresponding.Wherein can according to the match measure computing method of the embodiment of the present invention, calculate the match measure of two image blocks by what describe above.Particularly, can determine the match measure between the reference image block of reference picture and the object images piece of matching image by following operation: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
Utilization is determined method and match measure determining device according to the match measure of the embodiment of the present invention, in the match measure process of calculating pixel point or image block, consider the significance of pixel or image block, thereby can obtain, be more suitable for being applied to match measure that stereoscopic vision is mated, more objective, that more tally with the actual situation.
Utilization is according to the image matching method of the embodiment of the present invention and parallax calculation method, because in the match measure process of calculating pixel point or image block, considered the significance of pixel or image block, thereby can obtain, be more suitable for being applied to match measure that stereoscopic vision is mated, more objective, that more tally with the actual situation, and then can carry out more accurately images match and disparity computation.
Accompanying drawing explanation
Fig. 1 shows according to the overview flow chart of the match measure computing method towards stereoscopic vision coupling of first embodiment of the invention.
Fig. 2 has schematically shown based on pixel coupling and the difference based on image block match measure, and wherein (a) in Fig. 2 schematically shows respectively left and right figure; (b) in Fig. 2 schematically shows the match measure based on pixel; (c) in Fig. 2 schematically shows the match measure of cutting apart based on image.
Fig. 3 shows the schematic diagram that the match measure of traditional match measure and the above embodiment of the present invention is compared.
Fig. 4 illustrates according to the process flow diagram of the illustrative methods of the significance of definite image block of the embodiment of the present invention.
Fig. 5 illustrates the exemplary configuration block diagram of the disparity computation device mating towards stereoscopic vision according to an embodiment of the invention.
Fig. 6 shows according to the reference image block of reference picture towards stereoscopic vision coupling of the embodiment of the present invention and the functional configuration block diagram of the match measure determining device between the object images piece of matching image.
Fig. 7 illustrates the overall hardware block diagram of determining (or images match, or disparity computation) system according to the match measure of the embodiment of the present invention.
Embodiment
In order to make those skilled in the art understand better the present invention, below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
To be described in the following order:
1, invention thought general introduction
2, the first embodiment (match measure computing method)
2.1, match measure computing method example
2.2, determine the illustrative methods of the significance of image block
2.3, the match measure based on significance of the present invention on probability meaning solves
3, the second embodiment (image matching method)
4, the 3rd embodiment (parallax calculation method)
5, the exemplary configuration of disparity computation device
6, match measure determining device
7, system hardware configuration
8, sum up
1, invention thought general introduction
Sum up, stereoscopic vision coupling, for example the left and right images match in binocular stereo vision, can be divided into roughly local matching method and global registration method.
The basic thought of local matching method is according to certain similarity criterion, and relatively in two width images, the similarity of topography's piece determines that whether it is identical, thereby determines whether coupling.
Most of global registration algorithm is all based on the minimized thought of energy function, and target is to look for a parallax function, and global energy is minimized.As a kind of more conventional Global Algorithm, dynamic programming is for processing a kind of method of multistage decision problem, it is a kind of mathematical method that is used for reducing Global Optimal Problem complexity, it is undertaken by whole optimization problem is decomposed into decision-making stage by stage, whole cost function is by calculating stage by stage, conversion between each stage adds a series of constraints to carry out, for Stereo matching, the dull sequence constraint of outer polar curve allows overall coupling cost to realize by find minimal cost path on disparity space image, the cost of optimal path be all subpaths and, the coupling cost that these subpaths pass through a little can be decided by the similarity measurement factor of putting.Introduction about global registration algorithm, can be with reference to Daniel Scharstein, being entitled as of Richard Szeliski " A taxonomy and evaluation of dense two-frame stereo correspondence algorithms ", International Journal of Computer Vision, Volume 47Issue 1-3, April-June 2002, the article of Pages 7-42 and Michael Bleyer, Sylvie Chambon is entitled as " (3DPVT10) Does color really help in dense stereo matching " in International Symposium 3D Data Processing, the article of Visualization and Transmission (3DPVT) 2010.
As previously mentioned, pixel to or the right match measure of image block be for considering the basis of general image coupling or disparity computation.In the images match relation of overall disparity computation algorithm or the overall situation is evaluated, each pixel to or image block to match measure before, can be unified consideration, draw the overall the highest matching way of matching degree.
Inventor finds, in aforementioned prior art, when considering images match or disparity computation, as coupling each pixel of least unit or the status of each image block, is all identical, each pixel to or image between be completely independently.In other words, prior art calculating pixel to or during the right match measure of image block, only considered pixel to or image block to self, and ignored each pixel or the status of each image block in integral image, ignored in other words the relation of each pixel or each image block and integral image content.
Thus, inventor proposes, in stereoscopic vision coupling, especially for some overall parallax calculation methods, calculating pixel to or during the right match measure of image block, considered pixel or the image block status in integral image region or subregion is significance in other words, and design makes than prior art, the significance of pixel or image block is higher, and the match measure that relates to this pixel or image block is higher.
2, the first embodiment (match measure computing method)
2.1, match measure computing method example
Fig. 1 shows according to the overview flow chart of the match measure computing method 100 towards stereoscopic vision coupling of first embodiment of the invention.
Here supposition will be mated with reference to image and matching image, to determine that an image block (hereinafter referred to as reference image block) in reference image block for example mates with which in object images piece in matching image.About reference picture and matching image, for example, in the left image and right image of stereoscopic vision coupling, can an optional conduct with reference to image, another is matching image.
In addition, suppose that reference picture and matching image cut apart here.It is that the information such as imagery exploitation color and vein are divided into the operation in a plurality of regions to image that image in this paper meaning is cut apart, and each region will present similar color or texture.
Preferably, image is carried out to " over-segmentation " operation, compare with common cutting apart, over-segmentation refers to that the region area of cutting apart is as far as possible little, and number of regions is as far as possible many, makes in same cut zone, and parallax value equates.
Here, it should be noted that, as special case, the size of image block can be a pixel.
Fig. 2 has schematically shown based on pixel coupling and the difference based on image block match measure.(a) in Fig. 2 schematically shows respectively left and right figure; (b) in Fig. 2 schematically shows the match measure based on pixel.For pixel p in left figure, target is in right figure, to find corresponding p ', makes the match measure of p and p ' the highest; (c) in Fig. 2 schematically shows the match measure of cutting apart based on image.Target is in right figure, to find corresponding image block, make with left figure in comprise pixel p image block match.For instance, in the figure row alignment situation of left and right, if pixel p belongs to image block s in left figure, and in horizontal line L, all image blocks through horizontal line L in so right figure all will calculate match measure with image block s, and the parallax d that final image block s is corresponding can decide by the mean value of the parallax of all pixels in calculating s.
It should be noted that, block of pixels shown in Fig. 2 coupling or image block coupling carrying out in following situation: use mathematical method eliminated radially with tangential direction on lens distortion, adjusted the angle and distance between video camera, thereby obtained the correcting image that output row is aimed at, in the image of left and right, corresponding image block is positioned on this same level sweep trace.Thereby preferably, certain the image block A in given reference picture need to not search for the image block mating with image block A in the whole image of matching image, thus when doing match search, only need to be at the horizontal scanning line of matching image enterprising line search.
The method of the embodiment of the present invention can be applied to the coupling based on pixel, also can be applied to the coupling of image block.In following description, unless stated otherwise, the image block that image block is broad sense, the size that comprises image block is only the special circumstances of a pixel.
As shown in Figure 1, in step S110, the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated.
The method of the match measure between any computed image piece all can be estimated for the initial matching of calculating here.
As example, the initial matching of described computed image piece is estimated can use for example absolute difference and (Sum of Absolute Differences), the difference of two squares and (Sum of Squared Differences), normalization relevant (Normalized Cross-Correlation), aforementioned dynamic programming etc. method.
In step S110, obtain after initial matching estimates, advance to step S120.
In step S120, determine the significance of the reference image block of reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion.
Here, the significance of an image block can be for the general image of place, also can be for subregion.
For the example of subregion, as previously mentioned the in the situation that of the figure row alignment of left and right, when the reference image block in reference picture is done match search in matching image, can be only at the enterprising line search of par sweep trace of matching image.Thus, when determining the significance of an image block of reference picture, can be for region corresponding to the horizontal scanning line by this image block place.
Below, the method example of the significance of computing reference image block will be specifically described.
Determined the significance of reference image block in step S120 after, advance to step S130.
In step S130, the significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between the reference image block of reference picture and the object images piece of matching image.
The significance of hypothetical reference image block is S, and initial matching is estimated for R0, and the match measure after adjustment is R, and between three, relation can be expressed as formula (1):
R＝f(R0,S) …(1)
Wherein, function f meets in the situation that identical initial matching is estimated R0, and significance S is larger, makes the match measure R after adjusting larger.
For example, suppose that significance S span, for (0,1), can obtain the match measure after adjustment as follows
R＝R0*（1+S） …(2)
But, above formula (2) is only example, can adopt any functional form, as long as meet in the situation that same initial matching is estimated R0, the significance S of reference image block is larger, and the match measure R after adjustment is higher.
Fig. 3 shows the schematic diagram that the match measure of traditional match measure and the above embodiment of the present invention is compared.The left image that wherein in Fig. 3, represent to amplify in left side, in Fig. 3, right side represents in the situation of left and right figure row alignment the left image of comparing in prior art and the embodiment of the present invention and the pixel coupling signal of right image.As shown in Figure 3, in left figure, pixel n has the more pixel n ' 1 matching in right figure, n ' 2 and n ' 3 etc., and pixel m only only has a pixel m ' who matches.Main cause is that color, the texture of pixel n in image is all comparatively common, and pixel m shows as a lofty point, and in image, point is similarly few.Suppose that the colour-difference etc. of the wicket at pixel place calculates correlative factor at (m with match measure, m ') with (n, n ' 1) between all identical,, according to traditional match measure computing method, the match measure between pixel m and m ' equals or substantially equals the match measure between similitude n and pixel n ' 1; On the contrary, according to the match measure computing method of the embodiment of the present invention, if the definite match measure of the classic method of take is estimated as initial matching, the significance of pixel m and pixel n will be determined, then the traditional match measure separately of the significance adjustment based on separately, thereby because the significance of pixel m is significantly greater than the significance of pixel n, the pixel after adjusting also will be significantly higher than the match measure of pixel to <n-n ' 1> to the match measure of <m-m ' >.In fact, the confidence level of coupling <n-n ' 1> is obviously not mate the with a high credibility of <m-m ' >, therefore according to the match measure computing method of the embodiment of the present invention, this confidence level difference can be reflected at least to a certain extent, therefore according to the match measure method of the embodiment of the present invention, match measure more objective, that more tally with the actual situation can be obtained.
In above-described embodiment, the initial matching of only adjusting between this reference image block of obtained reference picture and the object images piece of matching image according to the significance of the reference image block of reference picture is estimated, but this is only example, the present invention is not limited thereto, alternatively, can also determine the significance of the object images piece of matching image, the significance of object images piece represents the significance degree of this object images piece in all image blocks of matching image; Wherein, initial matching between this reference image block of the reference picture that described adjustment obtains and the object images piece of matching image is estimated and is comprised, the initial matching that the significance of the significance of this reference image block based on this reference picture and this object images piece is adjusted between this reference image block of obtained reference picture and the object images piece of matching image is estimated, thereby determines the match measure between this reference image block of reference picture and the object images piece of matching image.
2.2, determine the illustrative methods of the significance of image block
Below with reference to Fig. 4, describe according to the illustrative methods of the significance of definite image block of the embodiment of the present invention.
Fig. 4 illustrates according to the process flow diagram of the illustrative methods 120 of the significance of definite image block of the embodiment of the present invention.This illustrative methods can be applied to the step S120 shown in Fig. 1.
As shown in Figure 4, in step S121, the number N of the Zone Full of statistical-reference image or the image block of subregion, wherein N is more than or equal to 2 integer.
In step S122, the number M of the image block similar to this reference image block in all image blocks of the Zone Full of computing reference image or subregion, wherein M is more than or equal to zero integer.
In statistics definite area, the number of the block similar to reference image block can comprise: for this reference image block in reference picture, calculate the similarity of other image blocks in the definite area of itself and this reference picture, can utilize color of image, shape, texture or moment characteristics to calculate here; Add up described similarity and be greater than the number of the image block of a fixed threshold, be the number of required similar block.
By the way, above moment characteristics is the concept in probability statistics, however in region and profile analysis also of great use.Can be from network address
http:// en.wikipedia.org/wiki/Image momentobtain the introduction about moment characteristics.
In step S123, the number M based on similar image piece, the number N of all images piece, the significance of computing reference image block, so that the number M of similar image piece is larger, the significance of reference image block is lower, and the number N of all images piece is larger, the significance of reference image block is lower.
For example, can determine by following formula the significance of described reference image block,
Wherein, K is greater than zero constant, and L is the constant that is greater than M, δ
1and δ
2for being greater than zero, be less than 1 real number, C is constant.Here " log " function is in order to prevent the appearance of very big or minimum numerical value, δ
1and δ
2be a balance parameters, object is in order to prevent that denominator from being 0, and L is a larger number, and for limiting the predetermined span of significance, for example L can be defined as the pixel summation of piece image.
But the method for the significance of this reference image block of above-mentioned definite reference picture is only example, the present invention is not limited thereto, any can be for determining that the method for the significance degree of an image block in the Zone Full of this image block place image or all image blocks of subregion all can be for the present invention.
For example, can be whether that one of angle point, statistical picture database or its combine the significance of determining this reference image block according to the gradient magnitude of reference image block, reference image block.
Particularly, for piece image, the place that gradient is larger and angle point region be edge or the most significant region of change color of image often, therefore the quantity of information comprising is larger, so can take the mode that detects angle point and compute gradient, assess significance, gradient magnitude is larger, and the detection degree of confidence of angle point is higher, and significance is larger.
In addition, staqtistical data base refers to sets up image data base in advance, for a certain image block in reference picture, by the number of image block similarly in computational data storehouse, carry out the significance of this piece image block in assessment reference image, similar image more at most significance is lower.
2.3, the match measure based on significance of the present invention on probability meaning solves
Below, in order to promote understanding, from probability meaning, explain solving of the match measure based on significance according to an embodiment of the invention.
Can from probability statistics angle by match measure regard as pixel to or image block to the degree matching (probability).Suppose do not considering under integral image ambient conditions, pixel to or image block degree that Q is matched or for probability P (Q) represent, the definite target of traditional match measure solves matching probability P (Q) exactly.In embodiments of the present invention, consider that match measure should be relevant to image overall content, suppose to be illustrated under the condition of two width integral image couplings with P (Q/I), pixel to or image block to the degree matching or probability, it is conditional probability, the subject object to be solved that is the embodiment of the present invention is solving condition probability P (Q/I), and I refers to picture material here, and P (I) represents the probability of two width images match.Theoretical according to naive Bayesian, we can access following formula (4),
Wherein, P (I/Q) be illustrated in pixel to or the condition of image block coupling under, the probability of two width images match, changes an angle and can be regarded as a pair of pixel or image block coupling for the contribution of two width images match.
Here, P (I) is constant for all pixels in image or image block, thus can draw following formula (5),
P(Q/I)∝P(Q)P(I/Q) …(5)
Symbol ∝ represents that its both sides are proportional.P (I/Q) can be regarded as a pair of pixel or image block mates the contribution for two width images match, can by the right significance of image block to be matched (S), be determined at least partly.By observing, can find, if the separating capacity that this block often appears in certain block in other words significance S just relatively a little less than, vice versa.Based on above-mentioned two factors, we can obtain,
P(I/Q)∝S …(6)
In conjunction with formula (5) and formula (6), we further can draw,
P(Q/I)∝P(Q)S …(7)
Therefore,, in order to calculate, need the significance S of calculating probability P (Q) and image block here.
As do not consider pixel under integral image ambient conditions to or image block degree or probability P (Q) that Q is matched, we can use a lot of traditional match measure, for example absolute differences and (Sum of Absolute Differences), the difference of two squares and (Sum of Squared Differences), normalization relevant (Normalized Cross-Correlation) etc. method.It should be noted that preferably these match measures are all normalized in the scope of 0-1, to meet the definition of probability.
About the calculating of significance S, can with reference to the aforesaid way that solves, for example, utilize formula (3) to solve.
In the situation that utilizing formula (3) to calculate significance, suppose to estimate R0 with traditional match measure or the so-called initial matching of this paper, P (Q), in conjunction with formula (7) and formula (3), can solve the match measure R after adjustment, i.e. P (Q/I)
3, the second embodiment (image matching method)
The image matching method for binocular stereo vision according to the embodiment of the present invention is described below, wherein applied describe above according to the match measure computing method of the embodiment of the present invention.
The image matching method for binocular stereo vision according to the embodiment of the present invention, can comprise the steps:
(1) input left image and right image;
(2) by left image and right image over-segmentation, be image block respectively;
(3) one of left image and right image are used as to reference picture, another,, as matching image, for the object images piece of the reference image block in reference picture and matching image, determines the match measure of the two:
(4) match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates,
Wherein can according to the match measure computing method of the embodiment of the present invention, calculate the match measure of two image blocks by what describe above.Particularly, can determine the match measure between the reference image block of reference picture and the object images piece of matching image by following operation: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
About the concrete operations of the match measure between computed image piece, can, with reference to the description to Fig. 4 in conjunction with Fig. 1 above, repeat no more here.
4, the 3rd embodiment (parallax calculation method)
The parallax calculation method for binocular stereo vision according to the embodiment of the present invention is described below, wherein applied describe above according to the match measure computing method of the embodiment of the present invention.
The parallax calculation method for binocular stereo vision according to the embodiment of the present invention, can comprise:
(1) input left image and right image;
(2) by left image and right image over-segmentation, be image block respectively;
(3) one of left image and right image are used as to reference picture, another,, as matching image, for the object images piece of the reference image block in reference picture and matching image, determines the match measure of the two:
(4) match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates;
(5) reference image block based in reference picture and the matching image piece in matching image, the parallax that computing reference image block is corresponding,
Wherein can according to the match measure computing method of the embodiment of the present invention, calculate the match measure of two image blocks by what describe above.Particularly, can determine the match measure between the reference image block of reference picture and the object images piece of matching image by following operation: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
About the concrete operations of the match measure between computed image piece, can, with reference to the description to Fig. 4 in conjunction with Fig. 1 above, repeat no more here.
About the reference image block based in reference picture and the matching image piece in matching image, the parallax that computing reference image block is corresponding, particularly, as example, can decide by the mean value of the parallax of all pixels in computing reference image block.
5, the exemplary configuration of disparity computation device
The exemplary configuration of the disparity computation device mating towards stereoscopic vision is according to an embodiment of the invention described below with reference to Fig. 5.
Fig. 5 illustrates the exemplary configuration block diagram of the disparity computation device 500 mating towards stereoscopic vision according to an embodiment of the invention.
As shown in Figure 5, disparity computation device 500 comprises: binocular camera 507, and by binocular camera 507, take and obtain left Figure 50 2 (1) and right Figure 50 2 (2); Image segmentating device 503, for cutting apart left figure and right Figure 50 2; The calculating of match measure can complete on chip 501, and chip 501 can or comprise statistics parts 504 and the match measure calculating unit 505 of cutting apart based on image for realization.In addition, disparity map calculation element 500 also comprises disparity map calculating unit 506, for based on matching degree tolerance result, calculates disparity map.The number of the number that the statistics parts 504 of cutting apart based on image of realizing or comprising in its chips 501 can be configured to statistical picture block and similar block.The number of for example adding up similar block comprises: for certain image block given in reference picture, calculate the similarity of other image blocks in Zone Full in itself and this reference picture or subregion, can utilize color of image, shape, texture or moment characteristics to calculate here; Add up described similarity and be greater than the number of the image block of a fixed threshold, be the number of required similar block.After trying to achieve the statistical information that image cuts apart, the method that match measure calculating unit 505 can for example be described referring to figs. 1 to Fig. 4 be carried out the match measure of computed image piece.
6, match measure determining device
Below with reference to Fig. 6, describe according to the reference image block of definite reference picture towards stereoscopic vision coupling of the embodiment of the present invention and the device of the match measure between the object images piece of matching image.
Fig. 6 shows according to the reference image block of reference picture towards stereoscopic vision coupling of the embodiment of the present invention and the functional configuration block diagram of the match measure determining device 600 between the object images piece of matching image.
As shown in Figure 6, match measure determining device 600 can comprise: initial matching is estimated acquisition parts 610, and the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Reference image block significance determining means 620, determines the significance of this reference image block of this reference picture, and the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; Match measure adjustment component 630, the significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
The concrete function that above-mentioned initial matching is estimated acquisition parts 610, reference image block significance determining means 620 and match measure adjustment component 630 can arrive the relevant description of Fig. 5 with Fig. 1 with reference to above-mentioned with operation.Here omit relevant being repeated in this description.
7, system hardware configuration
The present invention can also determine that by a kind of match measure (or images match, or disparity computation) hardware system implements.Fig. 7 illustrates the overall hardware block diagram of determining (or images match, or disparity computation) system 1000 according to the match measure of the embodiment of the present invention.As shown in Figure 7, match measure is determined (or images match, or disparity computation) 1000 can comprise: input equipment 1100, for input relevant image or information from outside, the parameter of the left image of taking such as video camera and right image, video camera etc., the remote input equipment that for example can comprise keyboard, Genius mouse and communication network and connect; Treatment facility 1200, for implementing the above-mentioned match measure according to the embodiment of the present invention, determine method, or be embodied as above-mentioned match measure determining device, or implement above-mentioned image matching method or parallax calculation method, what for example can comprise the central processing unit of computing machine or other has chip of processing power etc., can be connected to the network (not shown) such as the Internet, according to image after need to the processing to teletransmission of processing procedure etc.; Output device 1300, determines the result of (or images match, or disparity computation) process gained for implement above-mentioned match measure to outside output, for example, can comprise display, printer and communication network and the long-range output device that connects etc.; And memory device 1400, for storing the related data such as left and right image, image block number, significance, match measure, parallax of above-mentioned object detection (tracking) process in volatile or non-volatile mode, for example, can comprise the various volatile or nonvolatile memory of random-access memory (ram), ROM (read-only memory) (ROM), hard disk or semiconductor memory etc.
8, sum up
According to embodiments of the invention, a kind of reference image block of definite reference picture towards stereoscopic vision coupling and the method for the match measure between the object images piece of matching image are provided, wherein reference picture and matching image are divided into respectively image block, and this match measure determines that method can comprise: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
According to another embodiment of the present invention, a kind of reference image block of definite reference picture towards stereoscopic vision coupling and the device of the match measure between the object images piece of matching image are provided, wherein reference picture and matching image are divided into respectively image block, this match measure determining device can comprise: initial matching is estimated acquisition parts, and the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Reference image block significance determining means, determines the significance of this reference image block of this reference picture, and the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; Match measure adjustment component, the significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
According to an embodiment more of the present invention, a kind of image matching method for binocular stereo vision is provided, can comprise: (1) inputs left image and right image; (2) by left image and right image over-segmentation, be image block respectively; (3) one of left image and right image are used as to reference picture, another is as matching image, object images piece for the reference image block in reference picture and matching image, determine the match measure of the two: (4) match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates.Wherein can according to the match measure computing method of the embodiment of the present invention, calculate the match measure of two image blocks by what describe above.Particularly, can determine the match measure between the reference image block of reference picture and the object images piece of matching image by following operation: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
According to another embodiment of the invention, provide the parallax calculation method for binocular stereo vision according to the embodiment of the present invention, can comprise the steps: that (1) input left image and right image; (2) by left image and right image over-segmentation, be image block respectively; (3) one of left image and right image are used as to reference picture, another is as matching image, object images piece for the reference image block in reference picture and matching image, determine the match measure of the two: (4) match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates; (5) reference image block based in reference picture and the matching image piece in matching image, the parallax that computing reference image block is corresponding.Wherein can according to the match measure computing method of the embodiment of the present invention, calculate the match measure of two image blocks by what describe above.Particularly, can determine the match measure between the reference image block of reference picture and the object images piece of matching image by following operation: the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated; Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion; The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
Utilization is determined method and match measure determining device according to the match measure of the embodiment of the present invention, in the match measure process of calculating pixel point or image block, consider the significance of pixel or image block, thereby can obtain, be more suitable for being applied to match measure that stereoscopic vision is mated, more objective, that more tally with the actual situation.
Utilization is according to the image matching method of the embodiment of the present invention and parallax calculation method, because in the match measure process of calculating pixel point or image block, considered the significance of pixel or image block, thereby can obtain, be more suitable for being applied to match measure that stereoscopic vision is mated, more objective, that more tally with the actual situation, and then can carry out more accurately images match and disparity computation.
Aforementioned description is only illustrative, can much revise and/or replace.
In accompanying drawing above and description, with binocular stereo vision coupling, left and right images match is that example describes, but the present invention is not limited thereto, but can apply any images match for stereoscopic vision object.
In exemplary description above when determining the significance of pixel or image block, in the situation that significance is for for the image block in the subregion with respect to reference picture, take the horizontal region at this image block place is illustrated as example, but the present invention is not limited thereto, and this subregion can be the region that comprises the pre-sizing of be concerned about image block.
In exemplary description above, the match measure calculating represents be pixel to or the right matching degree of image block, wherein matching degree is higher, match measure is higher, but this is only example, it will be apparent to those skilled in the art that also and can estimate as match measure of the present invention by the conventional cost in this area, wherein two pixels or image block more do not mate more dissimilarly, and it is higher that corresponding cost is estimated (namely match measure).
Ultimate principle of the present invention has below been described in conjunction with specific embodiments, but, it is to be noted, for those of ordinary skill in the art, can understand whole or any steps or the parts of method and apparatus of the present invention, can be in the network of any calculation element (comprising processor, storage medium etc.) or calculation element, with hardware, firmware, software or their combination, realized, this is that those of ordinary skills use their basic programming skill just can realize in the situation that having read explanation of the present invention.
Therefore, object of the present invention can also realize by move a program or batch processing on any calculation element.Described calculation element can be known fexible unit.Therefore, object of the present invention also can be only by providing the program product that comprises the program code of realizing described method or device to realize.That is to say, such program product also forms the present invention, and the storage medium that stores such program product also forms the present invention.Obviously, described storage medium can be any storage medium developing in any known storage medium or future.
Also it is pointed out that in apparatus and method of the present invention, obviously, each parts or each step can decompose and/or reconfigure.These decomposition and/or reconfigure and should be considered as equivalents of the present invention.And, carry out the step of above-mentioned series of processes and can order naturally following the instructions carry out in chronological order, but do not need necessarily according to time sequencing, to carry out.Some step can walk abreast or carry out independently of one another.
Above-mentioned embodiment, does not form limiting the scope of the invention.Those skilled in the art should be understood that, depend on designing requirement and other factors, various modifications, combination, sub-portfolio can occur and substitute.Any modification of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection domain of the present invention.
Claims (10)
1. towards the reference image block of definite reference picture of stereoscopic vision coupling and a method for the match measure between the object images piece of matching image, wherein reference picture and matching image are divided into respectively image block, and this match measure determines that method comprises:
The initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated;
Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion;
The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
2. according to the match measure of claim 1, determine method, wherein, determine that the significance of this reference image block of this reference picture comprises:
The number N of the Zone Full of statistical-reference image or the image block of subregion, wherein N is more than or equal to 2 integer;
The number M of the image block similar to this reference image block in all image blocks of the Zone Full of computing reference image or subregion, wherein M is more than or equal to zero integer;
Number M based on similar image piece, the number N of all images piece, the significance of computing reference image block, so that the number M of similar image piece is larger, the significance of reference image block is lower, and the number N of all images piece is larger, and the significance of reference image block is lower.
3. according to the match measure of claim 2, determine method, wherein, by following formula, determine the significance of described reference image block,
Wherein, K is greater than zero constant, and L is the constant that is greater than M, δ
1and δ
2for being greater than zero, be less than 1 real number, C is constant.
4. according to the match measure of claim 1, determine method, wherein, determine that the significance of this reference image block of this reference picture comprises:
According to the gradient magnitude of this reference image block, reference image block, be whether that one of angle point, statistical picture database or its combine the significance of determining this reference image block.
5. according to the match measure of claim 1, determine method, wherein the size of image block is a pixel.
6. according to the match measure of claim 1, determine method, also comprise:
Determine the significance of the object images piece of this matching image, the significance of object images piece represents the significance degree of this object images piece in all image blocks of matching image;
Wherein, initial matching between this reference image block of the reference picture that described adjustment obtains and the object images piece of matching image is estimated and is comprised, the initial matching that the significance of the significance of this reference image block based on this reference picture and this object images piece is adjusted between this reference image block of obtained reference picture and the object images piece of matching image is estimated, thereby determines the match measure between this reference image block of reference picture and the object images piece of matching image.
7. according to the match measure of claim 1, determine method, wherein by " over-segmentation ", reference picture and matching image are carried out to cutting operation, so that in same cut zone, parallax value equates.
8. towards the reference image block of definite reference picture of stereoscopic vision coupling and a device for the match measure between the object images piece of matching image, wherein reference picture and matching image are divided into respectively image block, and this match measure determining device comprises:
Initial matching is estimated acquisition parts, and the initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated;
Reference image block significance determining means, determines the significance of this reference image block of this reference picture, and the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion;
Match measure adjustment component, the significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
9. for an image matching method for binocular stereo vision, comprising:
Input left image and right image;
By left image and right image over-segmentation, be image block respectively;
By one of left image and right image, as reference picture, another,, as matching image, for the object images piece of the reference image block in reference picture and matching image, determines the match measure of the two:
Match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates,
Wherein by following operation, determine the match measure between the reference image block of reference picture and the object images piece of matching image:
The initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated;
Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion;
The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
10. for a parallax calculation method for binocular stereo vision, comprising:
Input left image and right image;
By left image and right image over-segmentation, be image block respectively;
By one of left image and right image, as reference picture, another,, as matching image, for the object images piece of the reference image block in reference picture and matching image, determines the match measure of the two:
Match measure based on calculated, the matching image piece in the matching image that reference image block definite and in reference picture mates;
Reference image block based in reference picture and the matching image piece in matching image, the parallax that computing reference image block is corresponding,
Wherein by following operation, determine the match measure between the reference image block of reference picture and the object images piece of matching image:
The initial matching obtaining between this reference image block of reference picture and the object images piece of matching image is estimated;
Determine the significance of this reference image block of this reference picture, the significance of this reference image block represents the significance degree of this reference image block in the Zone Full of reference picture or all image blocks of subregion;
The significance of this reference image block based on this reference picture, this reference image block of reference picture that adjustment obtains and the initial matching between the object images piece of matching image are estimated, thereby determine the match measure between this reference image block of reference picture and the object images piece of matching image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210335757.7A CN103679686B (en) | 2012-09-11 | 2012-09-11 | Match measure method and apparatus, parallax calculation method, image matching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210335757.7A CN103679686B (en) | 2012-09-11 | 2012-09-11 | Match measure method and apparatus, parallax calculation method, image matching method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103679686A true CN103679686A (en) | 2014-03-26 |
| CN103679686B CN103679686B (en) | 2016-06-29 |
Family
ID=50317141
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210335757.7A Active CN103679686B (en) | 2012-09-11 | 2012-09-11 | Match measure method and apparatus, parallax calculation method, image matching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103679686B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104091339A (en) * | 2014-07-17 | 2014-10-08 | 清华大学深圳研究生院 | Rapid image three-dimensional matching method and device |
| CN104392448A (en) * | 2014-12-01 | 2015-03-04 | 四川大学 | Stereo matching method based on Gauss median segmentation guided filtering (GMSGF) |
| CN108074250A (en) * | 2016-11-10 | 2018-05-25 | 株式会社理光 | Matching power flow computational methods and device |
| CN108254738A (en) * | 2018-01-31 | 2018-07-06 | 沈阳上博智像科技有限公司 | Obstacle-avoidance warning method, device and storage medium |
| CN109461128A (en) * | 2018-10-24 | 2019-03-12 | 福州大学 | A kind of consistent stereo-picture color calibration method of structure kept based on gradient |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101739684A (en) * | 2009-12-17 | 2010-06-16 | 上海交通大学 | Color segmentation and pixel significance estimation-based parallax estimation method |
| EP2426930A1 (en) * | 2010-09-06 | 2012-03-07 | CSEM Centre Suisse d'Electronique et de Microtechnique SA - Recherche et Développement | System and method of analyzing a stereo video signal |
| CN102435174A (en) * | 2011-11-01 | 2012-05-02 | 清华大学 | Method and device for detecting barrier based on hybrid binocular vision |
-
2012
- 2012-09-11 CN CN201210335757.7A patent/CN103679686B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101739684A (en) * | 2009-12-17 | 2010-06-16 | 上海交通大学 | Color segmentation and pixel significance estimation-based parallax estimation method |
| EP2426930A1 (en) * | 2010-09-06 | 2012-03-07 | CSEM Centre Suisse d'Electronique et de Microtechnique SA - Recherche et Développement | System and method of analyzing a stereo video signal |
| CN102435174A (en) * | 2011-11-01 | 2012-05-02 | 清华大学 | Method and device for detecting barrier based on hybrid binocular vision |
Non-Patent Citations (3)
| Title |
|---|
| MIN-CHUL PARK等: "Automatic focusing attention for a stereo pair of image sequence", 《ICIS’09 PROCEEDINGS OF THE 2ND INTERNATIONAL CONFERENCE ON INTERACTION SCIENCES：INFORMATION TECHNOLOGY, CULTURE AND HUMAN》, 26 November 2009 (2009-11-26), pages 1185 - 1190 * |
| 曾志宏等: "融合深度信息的视觉注意计算模型", 《计算机工程》, vol. 36, no. 20, 31 October 2010 (2010-10-31), pages 200 - 202 * |
| 田明辉: "视觉注意机制建模及其应用研究", 《中国博士学位论文全文数据库信息科技辑(月刊)》, no. 09, 15 September 2010 (2010-09-15), pages 1 - 61 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104091339A (en) * | 2014-07-17 | 2014-10-08 | 清华大学深圳研究生院 | Rapid image three-dimensional matching method and device |
| CN104091339B (en) * | 2014-07-17 | 2017-01-11 | 清华大学深圳研究生院 | Rapid image three-dimensional matching method and device |
| CN104392448A (en) * | 2014-12-01 | 2015-03-04 | 四川大学 | Stereo matching method based on Gauss median segmentation guided filtering (GMSGF) |
| CN108074250A (en) * | 2016-11-10 | 2018-05-25 | 株式会社理光 | Matching power flow computational methods and device |
| CN108074250B (en) * | 2016-11-10 | 2022-01-04 | 株式会社理光 | Matching cost calculation method and device |
| CN108254738A (en) * | 2018-01-31 | 2018-07-06 | 沈阳上博智像科技有限公司 | Obstacle-avoidance warning method, device and storage medium |
| CN109461128A (en) * | 2018-10-24 | 2019-03-12 | 福州大学 | A kind of consistent stereo-picture color calibration method of structure kept based on gradient |
| CN109461128B (en) * | 2018-10-24 | 2021-09-28 | 福州大学 | Gradient-keeping-based color correction method for three-dimensional image with consistent structure |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103679686B (en) | 2016-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Campo et al. | Multimodal stereo vision system: 3D data extraction and algorithm evaluation | |
| US20120127275A1 (en) | Image processing method for determining depth information from at least two input images recorded with the aid of a stereo camera system | |
| JP6295645B2 (en) | Object detection method and object detection apparatus | |
| Correal et al. | Automatic expert system for 3D terrain reconstruction based on stereo vision and histogram matching | |
| US8897545B2 (en) | Apparatus and method for determining a confidence value of a disparity estimate | |
| CN103424105B (en) | Method for checking object and device | |
| Ma et al. | A modified census transform based on the neighborhood information for stereo matching algorithm | |
| CN104867133B (en) | A kind of quick substep solid matching method | |
| CN103679686A (en) | Match measure determination method, match measure determination device, parallax computation method and image matching method | |
| US8989481B2 (en) | Stereo matching device and method for determining concave block and convex block | |
| Saygili et al. | Hybrid kinect depth map refinement for transparent objects | |
| Kaczmarek | Improving depth maps of plants by using a set of five cameras | |
| Lladó et al. | Non-rigid metric reconstruction from perspective cameras | |
| Paudel et al. | 2D-3D camera fusion for visual odometry in outdoor environments | |
| CN110610503B (en) | Three-dimensional information recovery method for electric knife switch based on three-dimensional matching | |
| Harms et al. | Accuracy analysis of surface normal reconstruction in stereo vision | |
| Brink et al. | Probabilistic outlier removal for robust landmark identification in stereo vision based SLAM | |
| Xiong et al. | Color rank and census transforms using perceptual color contrast | |
| Cheung et al. | Optimization-based automatic parameter tuning for stereo vision | |
| CN108230382B (en) | Stereo matching algorithm based on Spearman correlation coefficient and dynamic programming fusion | |
| Basaru et al. | Quantized census for stereoscopic image matching | |
| Fleck et al. | Graph cut based panoramic 3D modeling and ground truth comparison with a mobile platform–The Wägele | |
| Gundewar et al. | Analysis of region based stereo matching algorithms | |
| CN115170745B (en) | Unmanned aerial vehicle distance measurement method based on stereoscopic vision | |
| CN114754732B (en) | Distance measurement method based on multi-eye vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |