CN105184848B - Deviation Control Method in Photon Mapping - Google Patents
Deviation Control Method in Photon Mapping Download PDFInfo
- Publication number
- CN105184848B CN105184848B CN201510498167.XA CN201510498167A CN105184848B CN 105184848 B CN105184848 B CN 105184848B CN 201510498167 A CN201510498167 A CN 201510498167A CN 105184848 B CN105184848 B CN 105184848B
- Authority
- CN
- China
- Prior art keywords
- photon
- collected
- joining
- carried out
- radiance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Image Generation (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to the Deviation Control Method in a kind of Photon Mapping, comprise the following steps：1) the estimation radius of joining caused by intersecting according to light with scene, is collected using KNN methods to the photon near joining；2) visibility judge is carried out to the photon being collected into, to reject impurity photon；3) detected by depth continuity and topological variation control is carried out to the photon being collected into；4) topological structure of surface normal is depended on according to the photon being collected into, geometric continuity detection is carried out with joining surface, so as to control impurity deviation；5) edge bias control is carried out to the photon being collected into by nonreentrant surface inspection；6) using remaining photon as correct and effective photon, radiance estimation is carried out using radiance calculation formula.Then obtained radiance calculated value is converted to color-values, you can synthesize final photo-realistic images.The present invention can effectively eliminate deviation and noise problem caused by incorrect collection photon.
Description
Technical field
The invention belongs to computer graphics techniques field, and in particular to the Deviation Control Method in a kind of Photon Mapping,
The photo-realistic images with good lighting effect can be drawn out using this method.
Background technology
People are endless, simulations for the demand that lighting effect true to nature is drawn out using computer graphics method
Complicated lighting effect is the great difficult problem that graphics researchers need to solve all the time in real world.Global illumination
(Global Illumination), the global illumination that is otherwise known as is bright, is the main way that photo rank photo-realistic images are achieved
Footpath, many algorithms, which are all attempted to provide for it, improves effective solution.What biggest advantage of light track algorithm and the present invention mainly studied
Photon Mapping algorithm is all the classic algorithm of global illumination, and compare ray tracing, Photon Mapping algorithm be recent two decades not
The disconnected Realistic Rendering technology to grow up.The research purpose of global illumination algorithm is gradually from drawing out effect true to nature at present
It is transferred on the demand of interactive application, the direction that the present invention is studied similarly provides important theory for the transfer of this demand
Foundation and practice reference.
Photon Mapping algorithm (Photon Mapping, abbreviation PM) is a kind of calculation that global illumination is realized based on photon figure
Method ((Jensen H.W.:Global illumination using photon maps.In Rendering Techniques
96.Springer, 1996, pp.21-30), photon figure is used for diffusing reflection component caused by computing environment, due to Photon Mapping phase
Caustic phenomenon and SDS phenomenons can be preferably simulated (caused by the propagation path of SDS difference acute pyogenic infection of finger tip light than ray tracing
Instead/refractive type, S are that mirror-reflection reflects specular, and D represents diffusing reflection diffuse, S and represents mirror-reflection/refraction
Specular, that is, refer to sequencing as illumination phenomenon caused by the propagation of minute surface-unrestrained-minute surface), the algorithm can obtain
Natural image more more life-like than ray tracing.
However, Photon Mapping algorithm is a kind of algorithm for having deviation and noise in itself.Radiance estimated accuracy seriously according to
Rely in the very small estimation kernel of radius and the accurate distribution of photons on estimation kernel and collection.It is inherently eliminated
Deviation error, it must just prevent the generation of visible deviation in radiance calculating, but all methods of forefathers include PPM
(Progressive Photon Mapping) and SPPM (Stochastic Progressive Photon Mapping) method
This requirement (Toshiya T., Ogaki S., Jensen H.W. can not be reached:Progressive photon
mapping.ACM Trans.Graph.27,5(Dec.2008),130:1–130:8；Toshiya T.,Jensen H.W.:
Stochastic progressive photon mapping.ACM Trans.Graph.28,5(Dec.2009),141:1–
141:8).Noise is another kind of wrong phenomenon caused by radiance estimation, caused by it main reason is that bright for radiating
The photon number that degree calculates is very few, and in the Photon Mapping algorithm of classics, noise is the threshold value that photon number is collected by setting
Come what is avoided.
Photon Mapping algorithm, which is one, inclined algorithm, and deviation Producing reason is that radiance estimation is substantially to calculate
Ray and the process of photon density at estimation surface joining x points, can only because x points are that volume and size is not present in abstract point
Approximately using the average photon density in x point near zones to replace the photon density of x points, so do will necessarily introduce it is not smart
True deviation.It is that can level off to correct result that although indirect illumination calculation levels off to infinity in photon number, real
On border, because the photon number of transmitting can not possibly reach infinitely great, the photon being distributed in radiance estimation region is typically
Limited even sparse, 0 can not be leveled off under conditions of limited photon by thus causing the radius of estimation kernel.Separately
Outside, also some other situation that may cause deviation occur：1) photon with joining distance " farther out " is also used in sometimes
The radiance of joining is calculated, this partial radiation amount is inaccurate；2) photon collection of mistake so that with currently intersecting
The photon of the no surface annexation of point is also collected into the radiance for calculating joining；3) photon of mistake is received
Collection so that there is surface to be connected with current joining and close on relation, but beyond the question on photon of the joining illumination without influence
It is collected into for calculating its radiance.Due to the above-mentioned feelings that may cause deviation be present in conventional Photon Mapping class method
Condition, its basic reason is that the collection of photon is in three dimensions kd-tree structures close on the search of photon, and photon
The geological information of accompanying partial 3 d scene surface is totally unknown, while is estimated in the range of the interior whole disk of verification
The hypothesis for having distribution of photons can not meet that our so caused errant radiation brightness estimation results are referred to as deviation in practice
Mistake.
Photon beam method can solve the problems, such as Photon Mapping algorithm edge bias (Havran V., Bittner J.,
Herzog R.,Seidel H.P.:Ray maps for global illumination.In Proceedings of the
Sixteenth Eurographics conference on Rendering Techniques(2005),Eurographics
Association,pp.43–54.).Based on Photon beam method, photon Splatting improves photon density estimation
(Herzog R.,Havran V.,Kinuwaki S.,Myszkowski K.,Seidel H.-P.:Global
illumination using photon ray splatting.In Computer Graphics Forum(2007),
Vol.26, Wiley Online Library, pp.503-513.), solve the insurmountable edge bias of Photon Mapping method
And the problem of topological variation.But photon is not launched and collected to Photon beam mapping algorithm, but directly transmitting and collection are penetrated
Line, therefore belong to two different class methods from Photon Mapping method.
The content of the invention
The present invention is directed to deviation and noise problem in the presence of Photon Mapping algorithm, there is provided inclined in a kind of Photon Mapping
Difference control method, it can effectively eliminate deviation and noise caused by incorrect collection photon.
To achieve the above object, the technical solution adopted by the present invention is as follows：
A kind of Deviation Control Method in Photon Mapping, comprises the following steps：
1) the estimation radius of joining caused by intersecting according to light with scene, using KNN methods to joining near
Photon is collected；
2) visibility judge is carried out to the photon being collected into, to reject impurity photon；
3) detected by depth continuity and topological variation control is carried out to the photon being collected into；
4) topological structure of surface normal is depended on according to the photon being collected into, geometric continuity is carried out with joining surface
Detection, so as to control impurity deviation；
5) edge bias control is carried out to the photon being collected into by nonreentrant surface inspection；
6) using remaining photon as correct and effective photon, carry out radiance using radiance calculation formula and estimate
Calculate.
Further, step 1) is received using the KNN methods based on kd-tree structures to the photon near joining
Collection, and photon number threshold value N is set to avoid the noise problem caused by collecting photon number deficiency.
Further, the comparison of surface normal and x point surface normals that step 2) is depended on according to the photon being collected into is entered
Row visibility judge.
Further, the method for step 3) the depth continuity detection is：Assuming that x points are corresponding on current imaging plane
The depth of pixel is dx, the depth for i-th of photon being collected into is diIf depth error threshold value is Δ d, then once | dx-di|>
Δ d, then current photon be removed, its corresponding area is also removed.
Further, the method for the continuity detection of the step 4) geometry is：
1) if the object ID where the ID and joining x of the object that photon i to be detected is depended on is entirely different, the light
Son does not pass through detection；
2) the body surface summit V nearest apart from joining x is foundx, find the body surface summit nearest apart from photon i
ViIf VxN ring neighborhood grid sets and ViN ring neighborhood grid sets between common factor be sky, or if VxN rings
Neighborhood vertex set NVn(Vx) and ViN ring neighborhood vertex sets NVn(Vi) between common factor be sky, then where showing photon
Surface does not pass through with the surface where x points without topological connection relation, the geometric continuity detection of the photon.
Further, step 5) by detect useful photon occupied area and calculate its area accumulation and, it is accurate to obtain
Estimation in long term voyage, realize the control to edge bias.
A kind of photo-realistic images method for drafting based on Photon Mapping, comprises the following steps：
1) photon tracing process is carried out, establishes photon figure；
2) ray tracing process is carried out, when light intersects with scene produces joining, is collected using the above method intersecting
Photon near point and the progress radiance calculating of the photon to being collected into；
3) obtained radiance calculated value is converted into color-values, synthesizes final photo-realistic images.
The present invention utilizes scene space thing where local surfaces distribution of photons by classifying to visible deviation Producing reason
Geometric properties possessed by body surface face, impurity photon rejecting, the control of impurity deviation, edge bias control and topology are carried out successively
Deviation controls, and the sequencing of above-mentioned deviation control can carry out free adjustment according to actual scene situation, have no strict priority
Sequentially.Using the above method, can eliminate estimating the visible deviation in kernel caused by non-precision photon collection, while can
To efficiently reduce caused noise error in radiance estimation.The main contributions of the inventive method can sum up effective elimination
Deviation and noise problem caused by incorrect collection photon.
Brief description of the drawings
Fig. 1 is the overview flow chart of the inventive method.
Fig. 2 is impurity photon schematic diagram, wherein (a) figure is side view, (b) figure is top view.
Fig. 3 is that typical scene carries out topological variation control exemplary plot, and the wherein upper left corner is scene graph to be drawn, and the lower left corner is
Partial enlarged drawing, right figure are the comparison figure of this method and other method drawing results.
Fig. 4 A are impurity deviation schematic diagram of a scenario.
Fig. 4 B are the comparison diagrams for carrying out the control of impurity deviation and not carrying out impurity deviation control, wherein (a) figure is the present invention
Method carries out the result figure of impurity deviation control, and (b) figure is the result figure for not carrying out impurity deviation control.
Fig. 4 C are that impurity deviation solves design sketch, wherein (a) figure is the result figure using the inventive method, (b) figure is to adopt
With the result figure obtained by conventional photonic mapping method.
Fig. 5 A are edge bias schematic diagrames.
Fig. 5 B are edge bias control contrast effect figures, wherein (a) figure is the result figure for not doing edge bias control, (b)
Figure is the result figure that edge bias control is carried out using the inventive method.
Embodiment
In order to facilitate the understanding of the purposes, features and advantages of the present invention, below by specific embodiment and
Accompanying drawing, the present invention will be further described.
The Photon Mapping of standard draws an image using twice of process, and first pass process establishes light using ray tracing
Subgraph (Photon Map), referred to as photon tracing process (Photon Tracing Pass)；Second time process is sent out from viewpoint position
Penetrate light to travel through in the scene, photon collection is just used according to the photon figure of storage when light intersects with scene produces joining
Method and the photon to being collected into carry out radiance calculating, and the radiance value for finally calculating joining is shown in image
On, referred to as ray tracing process (Ray Tracing Pass).Radiance calculating is carried out to the photon being collected into joining
Process be otherwise known as radiance estimation (Radiance Evaluation) or radiance estimation (Radiance
Estimate)。
The emphasis of the present invention is that the photon collection method during second time process i.e. ray tracing is innovated, and is proposed
A series of targetedly photon collections and filter method to ensure correctness and accuracy that final radiance calculates, from
And the deviation and noise phenomenon, acquisition in the photo-realistic images that can avoid or eliminate to generate as far as possible preferably draw effect.
Therefore, the present invention just for have novelty photon collection and filtration fraction be described, other parts due to other methods
Have no dramatically different, therefore here is omitted.
The flow chart of the inventive method is as shown in figure 1, comprise the following steps：
1) proceed by point x's using KNN algorithms according to the estimation radius of surface joining x in light and scene to be drawn
Photon collection；
2) visibility judge is carried out according to the comparison of the surface normal that the photon being collected into is depended on and x point surface normals,
Reject impurity photon；
3) detected by depth continuity and carry out topological variation control；
4) topological structure of surface normal is depended on according to the photon being collected into, geometric continuity inspection is carried out with x points surface
Survey, so as to control impurity deviation；
5) nonreentrant surface inspection is carried out, carries out edge bias control；
6) remaining photon carries out radiance using radiance calculation formula and estimated as correct and effective photon
Calculate.Then obtained radiance calculated value is converted to color-values, you can synthesize final photo-realistic images.
Illustrate the basic conception of the present invention first below, then illustrate above-mentioned each step.
The targeted rendered object of the present invention is the scene or model of three-dimensional, and three-dimensional scenic or model are in computer
Most usually using triangle gridding surface in figure and animation.The method of the present invention is applied to various polygonal mesh surfaces,
But in order to discuss conveniently, the present invention selects object of the most typical triangular mesh surface for discussion.The triangulation network is provided first
The related definition on lattice surface.Order：V={ v1,v2,…,vnRepresent triangle gridding vertex set, E={ el,e2,…,enRepresent triangle
Grid Edge collection, F={ fl,f2,…,fnTriangular topological relations collection is represented, M=(G, P) represents the triangulation network with two-dimensional manifold property
Lattice, wherein G=(V, E, F) represent to represent the topological relation and geometry between summit in set V by V, the connected graph of E, F generation, P
Information.
For arbitrary vi∈ M, define its 1 ring neighborhood (1-ring) vertex set NV1(i) it is：By with vertex viDirect phase
All summits even are formed；Define its 1 ring neighborhood (1-ring) tri patch set NT1(i)：By including vertex v1All three
Angular composition.Topological adjacency relation on surface mesh between summit is by NV1And NT (i)1(i) collectively constitute.According to opening up between summit
Relation is flutterred, can be with the N ring neighborhood vertex sets NV of defining pointnAnd N ring neighborhood tri patch set NT (i)n(i), N rings neighborhood
Vertex set NVn-thAnd N ring neighborhood tri patch set NT (i)n-th(i).| NV (i) | represent on grid in the Neighbourhood set of summit
Summit quantity, | NT (i) | represent tri patch Neighbourhood set intermediate cam piece number on grid.
Step 1：Collect kernel photon
For the ray and the joining x on surface on body surface, in order to calculate at joining x to ωiDirection (ωi
For incident direction) the radiance L that reflectss, needed to collect on photon figure near joining according to radiance estimation equation
Source of the photon as amount of radiation at joining.KNN (the k-Nearest based on kd-tree structures are used first
Neighbor) searching method is collected to the photon near x points.Estimated for ease of radiance and photon number threshold value N be set,
The amount of radiation of the nearest N number of photon of selected distance joining calculates radiance.Assuming that using joining x as the center of circle and include this N
The smallest circle radius surface of individual photon is r, then, joining near zone approximation is taken as the minimum disc on body surface, also known as
The disc is estimation kernel (Estimate Kernel), and radiance estimation is called interior nuclear estimation (Kernel again
Estimate).The purpose for setting photon number threshold value N is to avoid the noise problem caused by collecting photon number deficiency, calculated
Method needs photon to be stored in advance in support the spatial acceleration structure of KNN search such as to balance kd-tree (kd-tree is k-
The abbreviation of dimensional trees, it is a kind of data structure in segmentation k dimension datas space, is mainly used in hyperspace key number
According to search, such as range searching and nearest neighbor search, Kd-tree or kd-Tree can also be referred to as).Balancing kd-tree can be true
It is Ο (mlog (n)) to protect the time complexity for using kd-tree to position m photon in n photon.At this m photon
In the space closed on, finding the time of m photon in practice can further be reduced, so as to ensure the calculating of Photon Mapping algorithm
Efficiency.
Step 2：Reject impurity photon
Have collected in step 1 it is N number of close on photon, but not all r that is less than with joining x distances is (i.e. in intersecting
The spherical space of point x surfaces half) photon should all participate in radiance calculating.In this step, we use to the photon being collected into
The method of visibility judge, weed out to photon accompanying on the invisible dough sheets of current location x, while weed out the photon institute
Equivalent area in the estimation kernel occupied, the geometrical plane where so-called equivalent area refers to photon project to estimation kernel circle
Shared area on face.The method for the visibility judge that photon is depended between surface and x points have it is a variety of, it is most simple and be easy to real
The existing preferred comparison for carrying out normal direction of method, can be rejected in the photon with joining x backwards.Assuming that object table where photon p
The normal direction in face is(each photon is establishing the photon figure stage just while is recording the normal direction on the surface that the photon is depended on), and
Joining x normal direction isSo all satisfactionsPhoton can all be removed with remove radiance estimation in
Effect of errors.
As shown in Fig. 2 joining x, the disc that selection size is r on the section of the point are carried out on four water chestnut platforms in figure
Radiance estimates that left figure ((a) figure) is side view, and right figure ((b) figure) is top view, 1,2,3 three face Jun Bao of four water chestnut platforms
Containing distribution of photons, projection of the equivalent interior long term voyage caused by photon on estimation disc is respectively Δ on 1,2,3 face1, Δ2With
Δ3, wherein x points are located at Δ3Inside, Δ1And Δ2Respectively equivalent estimation kernel contributes the component of two crescent.By picking
Remove, finally only remain the photon on No. 3 faces, and the area of estimation kernel is
This step method solves may introduce mistake in conventional Photon Mapping method in some scenes well
Bleeding or light leakage phenomena.
Step 3：Topological variation controls
Previous step is rejected by impurity photon, has got rid of some impurity photons, but not remaining all photons
All it is effective, it is still desirable to continue to detect and judge its validity.As it is assumed that the photon for radiance estimation
The surface being all distributed in around joining x, the deviation error of radiance overestimation can be caused.This method passes through to being collected into
Photon carry out depth continuity detection, further to reject invalid photon.Assuming that x points correspond to picture on current imaging plane
The depth of element is dx, the depth for i-th of photon being collected into is diIf depth error threshold value is Δ d, then once | dx-di|>Δ
D, then current photon should be considered idler photon and be removed, its corresponding area is also removed.Key is Δ d setting, its
Value is the amount of an adaptive change, and different scenes use different values, and inside same scene, its value is also that can follow field
The change of scape position and change.
Fig. 3 illustrates the topological variation of scene in extreme conditions, and wherein x points are ray intersection point, and circle is photon
Radius r, the x point of collection are located on surface 2, and 1,3 represent other regions within circle radius r；The right is this hair in figure
Bright method and existing PM methods, the comparison figure of SPPM method drawing results, PM represent Photon Mapping methods, SPPM tables
Show Stochastic Progressive Photon Mapping methods.In scene, object opens up at the grid bar under desk
The distance that structure change is violent, is spaced closely together between every grid bar is flutterred, due to launching the limitation of photon number, photon kernel is estimated
The radius of calculation can not set it is too small be otherwise just possible to collect less than enough (i.e. N number of) photons.Estimate therefore, it is necessary to increase
Nuclear radius makes the estimation kernel of a joining even more be included comprising two on grid bar structures and ground in calculation
Photon, but larger estimation radius often introduces estimation error.By this step, positioned at earth's surface invalid photon by
In larger with the depth difference at grid bar surface x, so be removed.
Step 4：Impurity deviation controls
Even if the photon collected has passed through foregoing observability detection, depth detection, still there may be some impurity
Photon is invalid.Some photons be on the topological surface different from joining x (but with identical normal direction and approximately
Depth), this part has the photon radiation amount for splitting (discontinuous) topology therefore should be considered as idler photon and can not be by
Equally it is added in final radiance estimation.In this step, we carry out the control of impurity deviation, by photon institute according to
The geometric continuity detection in subordinate list face, and screened when radiance is estimated and weed out those and failed by the photon of detection
Method is made a return journey removal of impurity deviation.This partial radiation is abandoned when detecting that photon is not connecting (discontinuous) geometrically with joining
Amount, shone so as to avoid the result finally drawn from producing unnatural indirect light.
The continuity of geometry, which detects, is divided into two parts, and 1) if the ID (three-dimensional scenics for the object that photon i to be detected is depended on
The unique mark of middle object) it is entirely different with the object ID where joining x, then the photon will be considered nothing not by detection
Photon is imitated, detection terminates；Otherwise it is transferred to and detects in next step；2) the body surface summit V nearest apart from joining x is foundx, find
The body surface summit V nearest apart from photon iiIf VxN ring neighborhood triangle sets NTn(Vx) and ViN ring neighborhood triangles
Shape set NTn(Vi) between common factor for sky, i.e.,Then show the surface and x points place where photon
Surface without topological connection relation, the geometric continuity detection of the photon does not pass through, it should is removed as impurity photon；Together
Sample, geometric continuity detection can be carried out by the way of the common factor on neighborhood summit is judged, i.e., if VxN ring neighborhoods summit
Set NVn(Vx) and ViN ring neighborhood vertex sets NVn(Vi) between common factor for sky, namely
Surface where then showing photon does not lead to the surface where x points without topological connection relation, the geometric continuity detection of the photon
Cross, it should be removed as impurity photon.When realizing, the value of general n is 2 or 3.
Fig. 4 A are typical photon meeting all places in the scene that be still belonging respectively to two even more objects, figure
Photon in oval circle constitutes the photon set near x points.Mark on the desk of the left side is distributed in scene is
Photon is legal photon, and being distributed in the photon that the mark on the desk of the right side is will not have an impact to x points, and they should
This is taken as impurity photon to be deleted from radiance estimation.Because although they are in using x points as in the space sphere of the centre of sphere
Portion, plane where it are not connected with the grid surface where x points or in the presence of complete space obstacles, and the partial photonic should be by
It is considered as idler photon.If retaining such photon, such photon can produce deviation to final radiance estimation, actually
This kind of photon is also impurity deviation Producing reason, and the brightness effect either light leak that reflects that impurity deviation is usually expressed as mistake shows
As our method can be very good solve such problem.
Fig. 4 B are the comparison diagrams for carrying out the control of impurity deviation and not carrying out impurity deviation control, wherein (a) figure is the present invention
Method carries out the result figure of impurity deviation control, and (b) figure is the result figure for not carrying out impurity deviation control.Fig. 4 C are impurity deviations
Solves design sketch, wherein (a) figure is the result figure using the inventive method, (b) figure is using obtained by conventional photonic mapping method
The result figure with wrong lighting effect arrived.
Step 5：Edge bias controls
All it is effective photon mentioned by the remained photon of every test, but surface area occupied by photon
But might not all be effective, the calculating of its surface area is also not necessarily accurate.It is r discs by the center of circle and radius of joining x
Area be estimate kernel area, although abovementioned steps can be therewith the area occupied by idler photon when rejecting photon
Weed out simultaneously, but remaining area may also cause deviation as estimation kernel, the deviation is edge bias.Edge is inclined
Difference is a kind of due to bright in visible Low emissivity excessively caused by body surface marginal portion caused by nuclear estimation area in too high estimation
Degree estimation phenomenon, the basic reason of edge bias are photon caused by joining x is distributed around inequality.Intuitively, it is impossible to
Ensure to estimate photon and its amount of radiation deficiency using x will all to cause comprising photon in the whole disk in the center of circle.If using whole
Individual disk area calculates for radiance, then the image finally drawn will show unnatural dark and shade, so
And the photon in estimation kernel should be that joining x radiance estimation generation effective radiation is contributed.Such as
Shown in Fig. 5 A, effective estimation area at joining x should be included only such as the fan of lower left in the signified circle of white arrow
Shape region, but occur from other regions in circle and then searched for by KNN without useful photon is found, if still will not
Area comprising useful photon is also included in radiance estimation, then will necessarily influence the accuracy of final radiance estimation.
In the inventive method, edge bias control is carried out by nonreentrant surface inspection, i.e., by detecting face shared by useful photon
Product, and calculate its area accumulation and, so as to obtain accurately long term voyage in estimation, equivalent to having weeded out invalid kernel face
Product.The control for edge bias can finally be realized.Fig. 5 B are edge bias control contrast effect figures, wherein (a) figure is not
The result figure of edge bias control is done, (b) figure is the result figure that edge bias control is carried out using the inventive method.
Step 6：Radiance is estimated
By above-mentioned multiple detecting steps, using remaining photon as correctly useful photon, calculated using radiance
Formula carries out radiance estimation.Obtained radiance calculated value is converted to color-values, it is final true so as to synthesize
Feel image.
The above embodiments are merely illustrative of the technical solutions of the present invention rather than is limited, the ordinary skill of this area
Technical scheme can be modified by personnel or equivalent substitution, without departing from the spirit and scope of the present invention, this
The protection domain of invention should be to be defined described in claims.
Claims (8)
1. the Deviation Control Method in a kind of Photon Mapping, it is characterised in that comprise the following steps：
1) the estimation radius of joining caused by intersecting according to light with scene, the light near joining is collected using KNN methods
Son；
2) visibility judge is carried out to the photon being collected into, to reject impurity photon；
3) detected by depth continuity and topological variation control is carried out to the photon being collected into；
4) topological structure of surface normal is depended on according to the photon being collected into, geometric continuity inspection is carried out with joining surface
Survey, so as to control impurity deviation；
5) edge bias control is carried out to the photon being collected into by nonreentrant surface inspection；
6) using remaining photon as correct and effective photon, radiance estimation is carried out using radiance calculation formula.
2. the method as described in claim 1, it is characterised in that：Step 1) is used based on the KNN methods of kd-tree structures to phase
The photon of near intersections is collected, and sets photon number threshold value N to avoid the noise caused by collecting photon number deficiency
Problem.
3. the method as described in claim 1, it is characterised in that：The surface normal that step 2) is depended on according to the photon being collected into
Comparison with joining surface normal carries out visibility judge.
4. the method as described in claim 1, it is characterised in that the method for step 3) depth continuity detection is：Assuming that
The depth of joining respective pixel on current imaging plane is dx, the depth for i-th of photon being collected into is diIf depth is missed
Poor threshold value is Δ d, then once | dx-di|>Δ d, then current photon be removed, its corresponding area is also removed.
5. the method as described in claim 1, it is characterised in that the step 4) geometry continuity detection method be：
1) if the ID for the object that photon i to be detected is depended on and object ID where joining are entirely different, the photon is not
By detection, detection terminates；Otherwise it is transferred to and detects in next step；
2) the body surface summit V nearest apart from joining is foundx, find the body surface summit V nearest apart from photon ii, such as
Fruit VxN ring neighborhood grid sets and ViN ring neighborhood grid sets between common factor be sky, or if VxN ring neighborhoods top
Point set NVn(Vx) and ViN ring neighborhood vertex sets NVn(Vi) between common factor be sky, then show surface where photon with
Surface where joining does not pass through without topological connection relation, the geometric continuity detection of the photon.
6. the method as described in claim 1, it is characterised in that step 5) is by detecting useful photon occupied area and calculating it
The accumulation of area and, obtain accurately estimation in long term voyage, realize the control to edge bias.
7. the method as described in claim 1, it is characterised in that step 3)~5) execution sequence entered according to actual scene situation
Row adjustment.
8. a kind of photo-realistic images method for drafting based on Photon Mapping, it is characterised in that comprise the following steps：
1) photon tracing process is carried out, establishes photon figure；
2) ray tracing process is carried out, when light intersects with scene produces joining, using any one of claim 1~7
Methods described collects the photon near joining and the photon to being collected into carries out radiance calculating；
3) obtained radiance calculated value is converted into color-values, synthesizes final photo-realistic images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510498167.XA CN105184848B (en) | 2015-08-13 | 2015-08-13 | Deviation Control Method in Photon Mapping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510498167.XA CN105184848B (en) | 2015-08-13 | 2015-08-13 | Deviation Control Method in Photon Mapping |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105184848A CN105184848A (en) | 2015-12-23 |
| CN105184848B true CN105184848B (en) | 2018-01-09 |
Family
ID=54906898
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510498167.XA Active CN105184848B (en) | 2015-08-13 | 2015-08-13 | Deviation Control Method in Photon Mapping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105184848B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106777364A (en) * | 2017-01-22 | 2017-05-31 | 竹间智能科技（上海）有限公司 | Artificial intelligence response method and device that topic drives |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102981840A (en) * | 2012-11-08 | 2013-03-20 | 苏州两江科技有限公司 | Overall illumination computer simulation processing method based on photon mapping image |
| CN104200509A (en) * | 2014-08-19 | 2014-12-10 | 山东大学 | Photon mapping accelerating method based on point cache |
| CN104700448A (en) * | 2015-03-23 | 2015-06-10 | 山东大学 | Self adaption photon mapping optimization algorithm based on gradient |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB0902681D0 (en) * | 2009-02-18 | 2009-04-01 | Uws Ventures Ltd | Rendering caustics in computer graphics |
-
2015
- 2015-08-13 CN CN201510498167.XA patent/CN105184848B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102981840A (en) * | 2012-11-08 | 2013-03-20 | 苏州两江科技有限公司 | Overall illumination computer simulation processing method based on photon mapping image |
| CN104200509A (en) * | 2014-08-19 | 2014-12-10 | 山东大学 | Photon mapping accelerating method based on point cache |
| CN104700448A (en) * | 2015-03-23 | 2015-06-10 | 山东大学 | Self adaption photon mapping optimization algorithm based on gradient |
Non-Patent Citations (4)
| Title |
|---|
| Adaptive Progressive Photon Mapping;ANTON S.KAPLANYAN 等;《ACM Transactions on Graphics》;20130430;第32卷(第2期);16:1-16:13 * |
| Bias Compensation for Photon Maps;Roland Schregle;《COMPUTER GRAPHICS forum》;20040223;第22卷(第4期);729-742 * |
| 基于去噪光子图的光子映射;蔡鹏 等;《系统仿真学报》;20120908;第24卷(第9期);1844-1848 * |
| 基于自适应光子发射的渐进式光子映射;李睿 等;《计算机工程与设计》;20120116;第33卷(第1期);219-223 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105184848A (en) | 2015-12-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110120097B (en) | Semantic modeling method for airborne point cloud of large scene | |
| CN106127857B (en) | The on-board LiDAR data modeling method of integrated data driving and model-driven | |
| Gibbs et al. | Approaches to three-dimensional reconstruction of plant shoot topology and geometry | |
| Rottensteiner et al. | Results of the ISPRS benchmark on urban object detection and 3D building reconstruction | |
| Lafarge et al. | Creating large-scale city models from 3D-point clouds: a robust approach with hybrid representation | |
| Lafarge et al. | Building large urban environments from unstructured point data | |
| CN105488770B (en) | A kind of airborne laser radar point cloud filtering method of object-oriented | |
| Chauve et al. | Robust piecewise-planar 3D reconstruction and completion from large-scale unstructured point data | |
| CN102804231B (en) | Piecewise planar reconstruction of three-dimensional scenes | |
| CN106023312B (en) | Three-dimensional building object model automatic reconstruction method based on aviation LiDAR data | |
| CN109872397A (en) | A kind of three-dimensional rebuilding method of the airplane parts based on multi-view stereo vision | |
| CN109034077A (en) | A kind of three-dimensional point cloud labeling method and device based on Analysis On Multi-scale Features study | |
| CN109242862A (en) | A kind of real-time digital surface model generation method | |
| CN106504277B (en) | A kind of improved ICP point cloud autoegistration method | |
| CN109376900A (en) | Unmanned plane orbit generation method based on cloud | |
| CN107657659A (en) | The Manhattan construction method for automatic modeling of scanning three-dimensional point cloud is fitted based on cuboid | |
| CN110222642A (en) | A kind of planar architectural component point cloud contour extraction method based on global figure cluster | |
| WO2023124676A1 (en) | 3d model construction method, apparatus, and electronic device | |
| AU2010200144A1 (en) | Extraction processes | |
| CN113780389B (en) | Deep learning semi-supervised dense matching method and system based on consistency constraint | |
| Li et al. | New methodologies for precise building boundary extraction from LiDAR data and high resolution image | |
| Wang et al. | Oblique photogrammetry supporting procedural tree modeling in urban areas | |
| CN105184848B (en) | Deviation Control Method in Photon Mapping | |
| Han et al. | A method of the coverage ratio of street trees based on deep learning | |
| CN107993242A (en) | Based on airborne LiDAR point cloud shortage of data zone boundary extracting method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |