CN101615304A - Generate the method for the visual shell of robust - Google Patents

Abstract

A kind of method that generates the visual shell of robust belongs to field of three-dimension modeling.Comprise the steps: to obtain the image collection of object synchronization under at least two different points of view; Every width of cloth image in the described image collection is carried out parallel processing, obtain profile information and preservation; Utilize the line segment weighting to ask the friendship method to make up the initial segment aggregation model of object according to described profile information; Utilize the filtration of line segment aggregate central linear that the initial segment aggregation model is revised and obtain line segment aggregate model as a result; Utilize the line segment aggregate polygon to detect object is carried out resurfacing, obtain the visual shell of object.By the weighting line segment intersection, the line segment aggregate central linear is filtered the visual shell that the method that detects with the line segment aggregate polygon is calculated object, overcome traditional line segment and ask and give surface-boundary to detect inaccurate shortcoming, guarantee the degree of accuracy of arithmetic result simultaneously, strengthened the robustness of algorithm.

Description

Generate the method for the visual shell of robust

[technical field]

The present invention relates to a kind of method of three-dimensional modeling, especially relate to a kind of method that generates the visual shell of robust.

[background technology]

Obtaining the object dimensional Modeling Technology fast has a wide range of applications in fields such as computer vision, virtual reality, artificial intelligence, machine-building, digital entertainment and historical relic's protections.At present, three-dimensional model obtains technology and mainly contains and utilize modeling software structure three-dimensional model, obtain three-dimensional model and utilize image sequence to come these three kinds of methods of re-construct three-dimensional model by instrument and equipment.First method is utilized modeling software such as 3DMax, AutoCAD structure three-dimensional model, constructing virtual object and play up peculiar effect accurately, but that corresponding software uses is complicated, and the modeling cycle is long; Second method utilizes 3-D scanning equipment to obtain three-dimensional information, the modeling accuracy height, and the cycle is short, but equipment is expensive; The third method is according to the image sequence modeling, and the user only need use ordinary digital camera surrounding target object to take, and by photograph image information reconstruction model, its data acquisition equipment is simple, the efficient height.Therefore, based on the three-dimensional reconstruction of image sequence with its shirtsleeve operation, cheap equipment, higher efficient has become the important research object of computer vision.

Image sequence modeling based on two dimension is meant the technology of rebuilding the object master pattern by the silhouettes information on one group of 2-dimentional photo of object, and wherein the shooting angle of 2-dimentional photo is many more, and reconstructed results is approached real-world object more.Because the cone that produced of camera viewpoint and silhouettes intersects the visual shell that obtains object is included, so can only obtain the object model that approaches with real-world object.Algorithm based on the visual shell technology mainly is divided into two big classes at present: voxel method of cutting out and method of surface reconstruction.

Voxel method of cutting out, Martin and Aggarwal have proposed the voxel method that photo that camera from different perspectives photographs generates visual shell the earliest.Afterwards, Chien has proposed to utilize under parallel shooting condition octree structure to represent the method for visual shell, the voxel information of the data structure storage space cutting by Octree, this space expression formula can be handled cutting apart and canned data of space efficiently, has improved operation efficiency.R.Szeliski has put forward to improve algorithm at the Octree of angle shot arbitrarily under the basis of Chien.But on the whole, the method based on the voxel cutting has following shortcoming: the visual shell degree of accuracy that (1) reconstruction is come out is lower; (2) time complexity, space complexity is higher, even behind the storage organization of introducing Octree, problem still is not well solved; (3) surface is indirectly to be generated by separate voxel element, lack each other be closely connected and most situation under can produce redundant information.Along with the development of parallel multiprocessor technology, the voxel cut-out method can well be improved.

Method of surface reconstruction can solve the 3rd shortcoming of voxel method.Matsuik has proposed the method that polygon is asked friendship, and this algorithm directly asks intergrowth to become visual shell with the shooting viewpoint of camera with the formed cone of silhouettes line on the picture.Afterwards, Franco had used the direct reconstruction of objects of the method surface of how much of polar curves and Boundary Detection, had improved efficiency of algorithm.The work that this method comprised has: the polyhedron cap, the polygon cap detects the computing on limit.Go to the cutting object surface according to the side of each cone, directly write down the surface information of object.Though this method has solved the continuity between the surface, along with increasing of the quantity of taking pictures, the reconstructed surface of object can be very complicated.And the stability of most surfaces method for reconstructing all can be subjected to the influence of frontier point, causes reconstructed surface imperfect through regular meeting, especially when the object of rebuilding has complicated topological structure.In addition when the method for surface reconstruction modeling, also need to consider different constraint condition, for example smoothness constraint, block edge and texture edge, partial occlusion, reliability.

[summary of the invention]

In view of this, be necessary to provide a kind of method of rebuilding the generation visual shell of precision height, strong robustness.

A kind of method that generates the visual shell of robust comprises the steps: to obtain the image collection of object synchronization under at least two different points of view; Every width of cloth image in the described image collection is carried out parallel processing, obtain profile information and preservation; Utilize the line segment weighting to ask the friendship method to make up the initial segment aggregation model of object according to described profile information; Utilize the filtration of line segment aggregate central linear that the initial segment aggregation model is revised and obtain line segment aggregate model as a result; Utilize the line segment aggregate polygon to detect object is carried out resurfacing, obtain the visual shell of object.

By the weighting line segment intersection, the line segment aggregate central linear is filtered the visual shell that the method that detects with the line segment aggregate polygon is calculated object, overcome traditional line segment and ask and give surface-boundary to detect inaccurate shortcoming, the degree of accuracy of ensuring method result of calculation has strengthened robustness simultaneously.

Preferably, it is on the horizontal circumference at center that described viewpoint is uniformly distributed in the object, adopts digital camera shot object image under each viewpoint.

Preferably, adopt the profile information of the optimum ladder edge detection algorithm acquisition image of canny edge detection operator.

Preferably, the method for structure initial segment aggregation model comprises the steps:

To the profile of the object under each viewpoint, with its discrete set for point, the point in the described set all is positioned on the profile of object, constitutes dot profile, and each viewpoint and respective point profile constitute a viewpoint cone;

A bus of viewpoint cone and other viewpoint cones intersect generation busbar line segment collection, utilize the line segment weighting to ask the friendship method busbar line segment collection to be handled the result set that obtains under bus of this viewpoint cone;

All identical with the quantity of corresponding dot profile mid point in each viewpoint cone buses and other viewpoint cone are the initial segment aggregation model of object through the set of the result set after the processing identical with above-mentioned steps after crossing.

Preferably, described line segment weighting asks the friendship method to comprise the steps:

All line segments that busbar line segment is concentrated are represented with starting point and terminal point with one dimension parameter;

It is starting point that definition is concentrated minimum starting point with busbar line segment, first line segment that maximum terminal point is a terminal point belongs to the weighting line-segment sets, the busbar line segment collection comprises a plurality of sets of line segments, and each sets of line segments is by bus and crossing the obtaining of viewpoint cone, and the quantity of sets of line segments is identical with the quantity of other viewpoint cones;

Set an interim collection for empty;

Check whether each bar line segment and all line segments in the weighting line-segment sets in the sets of line segments are overlapping, if overlapping, overlapping result is inserted interim collection, and weighted count device that will be related with this overlapping result add one;

According to described interim collection the weighting line-segment sets is upgraded, and emptied interim collection;

Handle each sets of line segments one by one, concentrate all line segments all processed intact up to busbar line segment;

The line segment of choosing weighted count device maximum from final weighting line-segment sets constitutes result set.

Preferably, the step of renewal weighting line-segment sets comprises:

If it is overlapping that line segment in the weighting line-segment sets and the interim line segment of concentrating have, then the lap of this line segment is removed and formed new line segment, otherwise reservation;

The line segment aggregate of above-mentioned processing is gathered also with interim.

Preferably, resurfacing comprises the steps:

Line segment in the initial segment aggregation model is classified according to normal vector;

Situation according to the line segment coplane in the same class line segment obtains one group of plane, and each plane is distinguished according to the difference of intercept;

The crossing border that obtains of adjacent planar;

Adjacent boundary intersects to the limit, is configured for the polygon of resurfacing by the summit;

All line segments in the re-treatment initial segment aggregation model.

[description of drawings]

The synoptic diagram that Fig. 1 intersects for the viewpoint cone;

Fig. 2 is for generating the process flow diagram of visual shell;

Fig. 3 is for making up the process flow diagram of object initial segment aggregation model.

[embodiment]

In the present embodiment, the method that generates visual shell comprises the steps:

S1: the image collection that obtains object synchronization under at least two different points of view.This step is used to gather the object raw information of required modeling, and the viewpoint of collection multiple data quantity more is big more, and the reconstruction precision is high more, and it is 12 that present embodiment is gathered viewpoint.The image acquisition instrument of each viewpoint adopts common digital camera to get final product.It is on the horizontal circumference at center that each viewpoint is uniformly distributed in the object.Synchronization, object is taken from 12 different viewpoints, obtains an image collection.

S2: the every width of cloth image in the image collection is carried out parallel processing, obtain profile information and preservation.In this step, adopt edge detection algorithm to handle the profile information that obtains image.Edge of image is meant the part that the image local regional luminance is changed significantly, and the edge of image segment set has suffered the most information of image, determines that be very important with extracting the image border for the identification and the understanding of entire image scene.Present embodiment adopts the optimum ladder edge detection algorithm of canny edge detection operator to obtain the profile information of image.Specifically comprise:

S202: use the Gaussian filter smoothed image.

S204: with the finite difference of single order local derviation the assign to amplitude and the direction of compute gradient;

S206: gradient magnitude is carried out non-maximum value suppress;

S208: use the detection of dual threshold algorithm and be connected the edge.Obtain according to said method preserving behind the profile information of image, be used for further processing.

S3: utilize the line segment weighting to ask the friendship method to make up the initial segment aggregation model of object according to profile information.Detailed steps is as follows:

S302: obtain a profile,, constitute dot profile with the discrete set of profile for point.Each viewpoint constitutes a viewpoint cone with corresponding dot profile.

S304: the crossing busbar line segment collection that obtains of a bus of viewpoint cone and other viewpoint cones.Some viewpoint cones are crossing in bus and other viewpoint cones may obtain a point, also may obtain the line segment of a line segment or many non-overlapping copies.As shown in Figure 1, for viewpoint A being the synoptic diagram that some viewpoint cones intersect in the bus l of viewpoint cone on summit and other viewpoint cones.Bus l and this viewpoint cone intersect two line segment l of generation ₁, l ₂, form sets of line segments S ₁={ l ₁, l ₂.This bus l and the every other crossing busbar line segment collection { S that promptly generates under this bus l of viewpoint cone _i} _I=1 ^N, in the present embodiment, N=11, promptly bus l and other 11 viewpoint cones intersect, so the busbar line segment collection is { S ₁, S ₂..., S ₁₁, comprise 11 sets of line segments.

S306: utilize the line segment weighting to ask the friendship method busbar line segment collection to be handled the result set that obtains under bus of this viewpoint cone.Busbar line segment collection { S ₁, S ₂..., S ₁₁In, S ₁To S ₁₁Each other, utilize the line segment weighting to ask the friendship method that the busbar line segment collection is handled the result set that obtains under this bus l, it is to show organizing the maximum part of the overlapping each other number of times of all line segments that are on the same straight line that the core concept of friendship method is asked in the line segment weighting more.May have only a line segment in this result set, also have many line segments.

S308: result set is added the initial segment set.

S310: check from the point that viewpoint is guided into whether be last point of the processing on its loca profile.If, then be for further processing, if not, be back to S304.

S312: check whether handled profile is last profile.If, represent that then initial segment set interpolation finishes, if not, S302 then returned.

S314: obtain the initial segment aggregation model.

In the step of above-mentioned S306, exemplary algorithm is as follows:

With busbar line segment collection { S _i} _I=1 ^NIn all line segment represent with starting point and terminal point with one dimension parameter.Because busbar line segment concentrates all line segments all to be positioned on the bus, so its direction in the space is the same, so can only represent a line segment with starting point and terminal point with one dimension parameter.Wherein $S_i={\∪}_{k\∈\{\mathrm{1,2},...,m_i\}}[a_{\mathrm{ik}},b_{\mathrm{ik}}],$ a _ik≤b _ik≤a _ik+1。m _iExpression S _iThe bar number of middle conductor, a _Ik≤ b _Ik≤ a _Ik+1Expression is arranged in order according to the size of origin-to-destination numerical value.In present embodiment, S ₁={ l ₁, l ₂}=∪ _{K ∈ 1,2}}[a _1k, b _1k].

S602: find a _IkIn minimum value a ₀, b _IkIn maximal value b ₀With [a ₀, b ₀] expression first line segment, and this first line segment belongs to weighting line-segment sets S _OLSS, i.e. S _OLSS={ [a ₀, b ₀].Element count device M=1, record weighting line-segment sets S _OLSSIn element number.

S604: select a sets of line segments, as S ₁

S606: get a line segment in this sets of line segments, as [a ₁₁, b ₁₁].Setting interim collection S is empty set, promptly

S608: check institute's line taking section and weighting line-segment sets S among the S606 _OLSSIn all line segments whether have overlapping.The described overlapping starting point of certain bar line segment in two line segments and/or the terminal point of being meant is on another line segment.If overlapping, then overlapping result is inserted interim collection S, weighted count device that will be related with this overlapping result adds 1, and with weighting line-segment sets S _OLSSIn relate to overlapping line segment and remove also to insert behind this overlapping result and should collect S temporarily.As [a ₁₁, b ₁₁]=[2,4], [a ₀, b ₀]=[1,9], then overlapping result is [2,4], the weighted count device related with [2,4] adds 1.

S610: upgrade weighting line-segment sets S _OLSSWeighting line-segment sets S among the deletion S608 _OLSSIn relate to overlapping line segment, and will collect S and weighting line-segment sets S temporarily _OLSSMerge and obtain new weighting line-segment sets S _OLSSBar according to new weighting line-segment sets middle conductor is counted M, according to starting point sequential organization line segment [a from small to large ₀, b ₀], [a ₁, b ₁] ..., [a _M, b _M].As the result in the S608 example is S _OLSS={ [1,2], [2,4], [4,9] }, wherein the weighted count device numerical value of [2,4] is 1, the numerical value of the weighted count device of [1,2] and [4,9] is 0.

S612: check whether be the last item line segment in the sets of line segments.If not, then change step S606 over to; If then carry out next step.

S614: check whether be last sets of line segments.If not, then change step S604 over to; If then carry out next step.

S616: from the S as a result of above-mentioned processing _OLSSIn choose the maximum line segment of weighted count device and constitute result set.

The picture point profile comprises closely adjacent point, and the ray collection of the point on from the viewpoint to the dot profile is closely adjacent equally.Can calculate the result set of the crossing back gained of a ray and other cones effectively by the method for weighting line segment intersection, the set of the result set that all rays of dot profile are calculated should be the tight arrangement of rule, and can be subjected to The noise generation deviation by the approximate common factor result of the line segment aggregate in ray of method calculating of weighting line segment intersection, so be necessary it is revised.

S4: the initial segment aggregation model revised obtain line segment aggregate model as a result.This modification method is the center line filtration method, its to as if viewpoint under the set of all busbar line segment collection result set after treatment.

If { S _i} _I=1 ^N(N is the line segment number, S _iBe i bar line segment) be one group of line segment aggregate, and S _i=[a _i, b _i], a wherein _i≤ b _i, the central linear filtrator of this group line segment aggregate S ' _i} _I=1 ^N:

l _i=b _i-a _iExpression S _iLength

$l_i^{\′}={\mathrm{\Σ}}_{j=-N}^N{h}_j*l_{i+j},$ (wherein ${\mathrm{\Σ}}_{j=-N}^N{h}_j=1,$ N ∈ 3,5,7}); Adjusted length.

A ' _i=(b _i+ a _i)/2-l ' _i/ 2; The starting point that redefines.

B ' _i=(b _i+ a _i)/2+l ' _i/ 2; The terminal point that redefines.

If directly use $a_i^{\′}={\mathrm{\Σ}}_{j=-N}^N{h}_j*a_{i+j},$ And $b_i^{\′}={\mathrm{\Σ}}_{j=-N}^N{h}_j*b_{i+j}$ Filter, will occur the situation of the whole drift of line segment aggregate after level and smooth so.Line segment aggregate by the resulting visual shell model of weighting line segment intersection algorithm is the maximum approximation to the object visual shell, for the abnormal wire segment information that exists, the method that can filter by central linear is irregular regularization of line segment aggregate, make line segment aggregate overall trend trend obviously, be convenient to next step processing.

S5: utilizing as a result, the line segment aggregate model carries out resurfacing to object.If a light cone plane for a plurality of viewpoints as seen, then this light cone plane can intersect at the viewpoint cone under a plurality of visible viewpoints, obviously has only their common factor just can drop on the inside of final polyhedron visual shell.This light cone plane and other all photograph cones intersect the visual shell information that the polygon that obtains is object.Take above-mentioned principle to carry out resurfacing in the present embodiment, concrete steps comprise:

S502: the line segment in the initial segment aggregation model is classified according to normal vector, and promptly the line segment that normal vector is close is divided into a class, can set a threshold value in the application, as the close standard of determining method vector.

S504:, calculate the set of planes that this line segment aggregate produces to being in all line segments of same item.The line segment coplane that all normal vectors are identical constitutes a plane in this set of planes, and according to the intercept on plane the plane is distinguished.

S506: the plane that is adjacent is searched on each plane, the crossing border that obtains of adjacent planar.

S508: adjacent border is crossing to obtain the summit, and the summit connects the polygon that obtains being used for resurfacing in order.

S510: handle next similar line segment aggregate and all finish dealing with, can obtain the visual shell of object up to all line segments.

The above embodiment has only expressed one embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be interpreted as the restriction to claim of the present invention.Should be pointed out that for the person of ordinary skill of the art without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.

Claims (7)

1, a kind of method that generates the visual shell of robust is characterized in that, comprises the steps:

Obtain the image collection of object synchronization under at least two different points of view;

Every width of cloth image in the described image collection is carried out parallel processing, obtain profile information and preservation;

Utilize the line segment weighting to ask the friendship method to make up the initial segment aggregation model of object according to described profile information;

Utilize the filtration of line segment aggregate central linear that the initial segment aggregation model is revised and obtain line segment aggregate model as a result;

Utilize the line segment aggregate polygon to detect object is carried out resurfacing, obtain the visual shell of object.

2, the method for the visual shell of generation robust as claimed in claim 1 is characterized in that, it is on the horizontal circumference at center that described viewpoint is uniformly distributed in the object, adopts digital camera shot object image under each viewpoint.

3, the method for the visual shell of generation robust as claimed in claim 1 is characterized in that, adopts the profile information of the optimum ladder edge detection algorithm acquisition image of canny edge detection operator.

4, the method for the visual shell of generation robust as claimed in claim 1 is characterized in that, the method that makes up the initial segment aggregation model comprises the steps:

To the profile of the object under each viewpoint, with its discrete set for point, the point in the described set all is positioned on the profile of object, constitutes dot profile, and each viewpoint and respective point profile constitute a viewpoint cone;

A bus of viewpoint cone and other viewpoint cones intersect generation busbar line segment collection, utilize the line segment weighting to ask the friendship method busbar line segment collection to be handled the result set that obtains under bus of this viewpoint cone;

All identical with the quantity of corresponding dot profile mid point in each viewpoint cone buses and other viewpoint cone are the initial segment aggregation model of object through the set of the result set after the processing identical with above-mentioned steps after crossing.

5, the method for the visual shell of generation robust as claimed in claim 4 is characterized in that, described line segment weighting asks the friendship method to comprise the steps:

All line segments that busbar line segment is concentrated are represented with starting point and terminal point with one dimension parameter;

It is starting point that definition is concentrated minimum starting point with busbar line segment, first line segment that maximum terminal point is a terminal point belongs to the weighting line-segment sets, the busbar line segment collection comprises a plurality of sets of line segments, and each sets of line segments is by bus and crossing the obtaining of viewpoint cone, and the quantity of sets of line segments is identical with the quantity of other viewpoint cones;

Set an interim collection for empty;

Check whether each bar line segment and all line segments in the weighting line-segment sets in the sets of line segments are overlapping, if overlapping, overlapping result is inserted interim collection, and weighted count device that will be related with this overlapping result add one;

According to described interim collection the weighting line-segment sets is upgraded, and emptied interim collection;

Handle each sets of line segments one by one, concentrate all line segments all processed intact up to busbar line segment;

The line segment of choosing weighted count device maximum from final weighting line-segment sets constitutes result set.

6, the method for the visual shell of generation robust as claimed in claim 5 is characterized in that, the step of upgrading the weighting line-segment sets comprises:

If it is overlapping that line segment in the weighting line-segment sets and the interim line segment of concentrating have, then the lap of this line segment is removed and formed new line segment, otherwise reservation;

The line segment aggregate of above-mentioned processing is gathered also with interim.

7, the method for the visual shell of generation robust as claimed in claim 1 is characterized in that, resurfacing comprises the steps:

Line segment in the initial segment aggregation model is classified according to normal vector;

Situation according to the line segment coplane in the same class line segment obtains one group of plane, and each plane is distinguished according to the difference of intercept;

The crossing border that obtains of adjacent planar;

Adjacent boundary intersects to the limit, is configured for the polygon of resurfacing by the summit;

All line segments in the re-treatment initial segment aggregation model.

Images