CN101398940A - True object material quality reflecting attribute modeling approach based on linear light - Google Patents

Abstract

A method for modeling a reflection attribute of a real object material based on a linear light source comprises the following steps: (1) a data collection device based on the linear light source collects and process data; the linear light source is taken as a light source system of the data collection device; the linear light emitted by the linear light source shines a collected plane object, a control system drives an automatic translation system to move forwards; meanwhile, an industrial camera collects images of the collected plane object, a computer performs subsequent processing of the collected image data to obtain a trajectory curve of pixel values and store the trajectory curve; (2) a reflection table is generated; (3) BRDF attribute is modeled to obtain rhod of a Ward illumination model, and then obtain alpha, and finally obtain rhos according to the trajectory curve of the pixel values obtained in the step (1) and the reflection generated in the step (2), wherein, the rhod is a diffuse reflection parameter, the rhos is a mirror reflection parameter and the alpha is a material roughness coefficient. The method can efficiently collect data and quickly solve the parameters of the BRDF model, and is easy to be realized.

Description

True object material quality reflecting attribute modeling approach based on linear light sorurce

Technical field

The invention belongs to the computer virtual reality technology field, specifically a kind of flat object material reflecting attribute modeling method of utilizing linear light sorurce is used for modeling and the drafting of realistic virtual environment material BRDF.

Background technology

The real-world object three-dimensional modeling is the pith of computer graphics.Three-dimensional modeling is divided into Geometric Modeling and attribute modeling, and Geometric Modeling can be by fine realizations such as modeling software and spatial digitizers, and the attribute modeling then becomes the difficult point of a computer graphics modeling.The accurate collection of real-world object surface reflection attribute and the modeling hot issue that is computer graphics wherein.The outward appearance of object is the result by the material effect of light and object.How the reflecting attribute modeling interacts in order to describe light and body surface exactly.The model of at present the most frequently used description reflecting attribute is bidirectional reflectance distribution function BRDF (Bi-directional Reflectance Distribution Function).Document 1-Nicodemus, F.E., Richmond, J.C., Hsia, J.J., Ginsberg, I.W., and Limperis, among the T.Geometric considerations andnomenclature for reflectance.National Bureau of Standards Monograph 160,1977 at first complete description the BRDF definition, it has described the incident light that arrives body surface and ratio from the emergent light of this position reflection.If fixedly incident light and emergent light direction are rotated the material sample around surface normal, the BRDF value remains unchanged, and then this type of material is the isotropy material; Otherwise then be called the anisotropy material.BRDF meets physical ultimate principle, has non-negative, energy conservation and reversible key properties such as (Reciprocity).In the real world, many objects are made of many materials, need BRDF (Spatially Varying BRDF, SVBRDF) description body surface reflecting attribute with spatial variations for this reason.

The BRDF data acquisition equipment is early stage collecting device based on angle measurement reflectometer (gonioreflectometer), as document 2-K.Torrance, E.Sparrow, and R.Birkebak, " Polarization; directional distribution; andoff-specular peak phenomena in light reflected from roughened surfaces; " Optical Society ofAmerica, vol.56, no.7, pp.916-925,1966 and document 3-F.Sing-Choong, " A Gonioreflectometerfor Measuring the Bidirectional Reflectance of Material for Use in Illumination Computation, " Computer Science, Cornell University, the equipment of using in 1997, the precision of such collecting device is higher, but its whole image data process is consuming time relatively.The collecting device that is based on digital camera of present commonplace application.The two dimensional image of camera collection can obtain a large amount of BRDF data, this kind equipment can be gathered the BRDF data of single material, but relatively be subjected to be gathered the restriction of material, not general, as document 4-W.Matusik, P.Hanspeter, B.Matthewet al., " Efficient isotropic BRDF measurement; " in Proceedings of the 14th Eurographicsworkshop on Rendering, Leuven, Belgium, pp.241-247, the equipment of using in 2003, can also gather the SVBRDF data of many materials, document 5-K.Dana, J., the equipment of using among the ＂ BRDF/BTF measurement device. ＂ pp.460-466, the collecting device that can gather SVBRDF fast uses parabolic mirror to replace complicated mechanical to control the scope that enlarges into diapason and observation domain, and above-mentioned collecting device cost is too high, poor efficiency consuming time.

Next data for obtaining through collecting device enter the BRDF attribute modelling phase usually, come data are handled to obtain mathematical model or empirical model.The modeling method of BRDF mainly contains two big classes: a class is based on the modeling method of parametrization mathematic(al) representation, by the material sample is carried out sparse sampling, the BRDF model parameter of the known analytic expression of Data Matching that use is obtained, in drawing process,, analytic expression carries out the reflex simulation by being calculated in real time, document 6-B.T.Phong for example, Illumination for computer generated pictures, Communications ofACM vol.18, no.6, pp311-317, use the Phong model of the approximate direct reflection of simple cosine lobe (cosine lobe) in 1975, document 7-E.Lafortune, S.-C.Foo, etc.Non-linear approximation of reflectancefunctions, Computer Graphics, 1997, no.Annual Conference Series:pp.117-126 is improved on Phong model basis, Lafortune model physically more true to nature and document 8-Ward, G.J.Measuring andmodeling anisotropic reflection.In SIGGRAPH, 1992, use Gauss to replace the Ward model of cosine among the pp.265-272; The another kind of BRDF modeling method that is based on data-driven (Data-Driven), as 4 pairs of material intensive samplings of document, use linearity and nonlinear analysis, generate new BRDF data by interpolation and Extrapolation method, and provide some user interactions of parameter confession more intuitively to revise BRDF, this class methods modeling complexity needs to gather mass data.

Summary of the invention

Technology of the present invention is dealt with problems: for overcoming above-mentioned shortcoming, the invention provides a kind of true object material quality reflecting attribute modeling approach based on linear light sorurce, this method has made up the collecting device of a cover based on linear light sorurce, image data is found the solution the BRDF model parameter fast efficiently.

The present invention adopts the true object material quality reflecting attribute modeling approach based on linear light sorurce, and its characteristics are that step is as follows:

(1) based on the data acquisition equipment of linear light sorurce to data acquisition and processing (DAP)

The data acquisition equipment of described linear light sorurce comprises:

The Le Gao robot is as the control system and the automatic translation system of data acquisition equipment;

Industrial camera is as the view data capture systems;

Computing machine is as the data handling system of data acquisition equipment;

Linear light sorurce is as the light-source system of data acquisition equipment;

Linear light sorurce sends linear light irradiation on the quilt planar object of gathering, control system drives automatic translation system and moves forward, industrial camera carries out image acquisition to the plane of being gathered simultaneously, and computing machine carries out follow-up data to the view data of gathering to be handled, and obtains pixel value geometric locus Image _{U, v}(x) also storage, wherein x represents the current images data directory, and the scope of x is [1, n], and n is the image number, and u and v represent the position of current pixel point in image;

(2) utilize linear light sorurce in the step (1), generate the reflection table;

(3) BRDF attribute modeling;

According to pixel value geometric locus and the middle reflection table that generates of step (2) that step (1) obtains, obtain the ρ of Ward illumination model _d, and then obtain α, obtain ρ at last _s, ρ wherein _dThe diffuse reflection parameter, ρ _sThe direct reflection parameter, α is the material coefficient of roughness.

The method that described step (1) obtains the pixel value geometric locus is: is a series of image definition data Image _{U, v}(x), wherein x represents the current images data directory, and the scope of x is [1, n], and n is the image number, and u and v represent the position of current pixel point in image.

Generate the step of reflection chart picture in the described step (2):

(a) suppose diffuse reflection parameter ρ in the Ward model _d=1, partly utilize linear light sorurce to carry out light source linearization integration for diffuse reflection, draw out diffuse reflection table D (x);

(b) suppose direct reflection parameter ρ in the Ward model _s=1, partly utilize linear light sorurce to carry out light source linearization integration for direct reflection, draw out direct reflection table S _α(x), the α material coefficient of roughness wherein, its span is 0-0.3;

(c) for each current images data directory x, ask mathematical expectation μ and variances sigma;

(d) to step (a) and two reflection tables that generate (b) be optimized, it is as follows to optimize formula:

S _α' (x)=and 00＜α '＜0.005, α ' is the material coefficient of roughness after optimizing.

Described step (3) diffuse reflection parameter ρ _dObtain as follows:

ρ _dValue as shown in the formula:

${\mathrm{\ρ}}_d={\mathrm{\ρ}}_{s\tan \mathrm{dard}}\·\frac{L_d}{L_{s\tan \mathrm{dard}}}$

ρ _StandardBe known diffuse reflection parameter, L _StandardBe known diffuse reflection parameter ρ _StandardReflection peak, L _dThe diffuse reflection peak value of the pixel value geometric locus that obtains after handling for computer data;

The method that obtains of the material coefficient of roughness α of described step (3) is: when linear light sorurce runs to direct reflection direction x _mDuring the place, the direct reflection peak value of this point produces, and according to knowledge of statistics, obtains the mathematical expectation μ and the variances sigma of pixel value on this direct reflection geometric locus, and all pixel value sum S, promptly tries to achieve according to following three formulas:

$\mathrm{\μ}=v\frac{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}x\·\mathrm{Image}\left(x\right)}{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)}$

${\mathrm{\σ}}^2=v\frac{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}{(x-\mathrm{\μ})}^2\·\mathrm{Image}\left(x\right)}{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)}$

$S=v\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)$

X wherein _mBe the position at direct reflection peak value place,, x _dBe diffuse reflection peak value position, Image (x) is the x width of cloth view data of gathering, and gets with respect to x _mPixel value in the zone of symmetry is promptly from x _m=2x _m-x _dTo x _d, because camera is to the influence of this regional projective transformation, the width that different pixels is corresponding different, so all need scaledly for all values, zoom factor is a following formula:

$v=\frac{1}{2(x_d-x_m)}$

After obtaining these statistical sample data, from the reflection table of the linear light sorurce drawn out before, the σ in the corresponding reflection table of σ that obtains calculating with statistical sample of tabling look-up is again because σ and α are one to one in the linear light sorurce reflection table, so just can obtain α, and unique.

Described step (3) direct reflection parameter ρ _sFormula:

${\mathrm{\ρ}}_s={\mathrm{\ρ}\′}_{s\tan \mathrm{dard}}\·\frac{\mathrm{Ls}}{{L\′}_{s\tan \mathrm{dard}}}$

Wherein: ρ ' _StandardBe known minute surface diffuse reflection parameter, L ' _StandardBe known minute surface diffuse reflection parameter ρ ' _StandardReflection peak, Ls is the direct reflection peak value of the pixel value geometric locus that obtains after computer data is handled.

The present invention's advantage compared with prior art is: made up the efficiently collecting device of a cover based on linear light sorurce, only needed to gather hundreds of group data image, the material properties data of all directions that just can the acquisition plane real-world object; Utilize linear light sorurce and reflection table to find the solution the BRDF model parameter fast, avoid utilizing that the Nonlinear Numerical optimization method finds the solution is consuming time, calculated amount big and the numerical value instability problem.

Description of drawings

Fig. 1 is one-piece construction figure of the present invention;

Fig. 2 is the data acquisition equipment structural representation based on linear light sorurce of the present invention;

Fig. 3 is a light source linearization reflection table of the present invention;

Fig. 4-a, Fig. 4-b, Fig. 4-c are three parameters of Ward model of the present invention, and Fig. 4-a is the parameter ρ of the Ward model of trying to achieve _d, Fig. 4-b is the parameter ρ of the Ward model of trying to achieve _s' Fig. 4-c is the parameter alpha of the Ward model of trying to achieve;

Fig. 5-a, Fig. 5-b are experimental result of the present invention, and Fig. 5-a is the true picture of object, and Fig. 5-b is the drafting design sketch of object.

Embodiment

The present invention is primarily aimed at the BRDF modeling of smooth real-world object Facing material reflecting attribute.

As shown in Figure 1, specific implementation process of the present invention comprises three key steps: data acquisition and processing, BRDF modeling and GPU real-time rendering.

Step 1: data acquisition and processing

Data acquisition and treatment step are the bases of subsequent step, comprise as the next stage:

(1) data acquisition equipment builds

As shown in Figure 2, the collecting device of the present invention's employing is the linear light sorurce data acquisition equipment.The halogen spot that linear light sorurce device wherein adopts automobile decoration to use, 50 centimeter length, diameter is 2 centimetres, with the not reflective adhesive plaster parcel of black, make a side of light source become long 50 centimetres, wide about 3 millimeters linear light sorurce, the assurance width can be enough narrow, ignores to be theoretic linear light sorurce.

Use mental storm (the LEGO MindStroms Robotics) NXT#8527 of Le Gao robot as control system and automatic translation system, and make the two-wheel of carrying linear light sorurce of its accessory, two-wheel below track is made of the U type aluminium groove of 1 meter long of two treaty and 1 centimetre wide, automatically translation system one side hangs a weight down with fishing line, the opposing party then is connected on the rotating pulley of the #8527 of Le Gao robot with fishing line, and translation system can be controlled moving horizontally of linear light sorurce automatically.Automatically translation system is fixed on linear light sorurce from the about 5 centimetres height of being gathered of object plane, and it is at the uniform velocity parallel mobile to draw linear light sorurce.

Industrial camera is made up of industrial camera DH-SV1410 of Daheng and camera control program, and camera is placed on the upwards normal vector of acquisition plane to become on the direction at 60 degree angles, is oriented acquisition plane.

(2) collection of BRDF data

At first the space is gathered in definition, promptly acquisition plane is treated as by X-axis and the well-determined plane of Z axle, and linear light sorurce is parallel to the Z axle, prolonging the direction parallel with X-axis moves, the acquisition plane center is an initial point, and the distant place of camera is to being the X-axis positive dirction, and Y-axis is the three dimensions of positive dirction vertically upward.

In the actual acquisition, the planar object of needs collection is put on the collecting device desktop, flattens both sides with the mute light bonding jumper of black.Camera frame to plane X=0 a bit, camera point becomes vector and Y-axis forward angle about 60 to spend with initial point, this angle will guarantee that data that this camera position collects enough separate diffuse reflection peak value and direct reflection peak value in the brightness value track of pixel.

API by the #8527 of Le Gao robot carries out the parameter setting to it then, makes it can at the uniform velocity prolong X-axis forward traction linear light sorurce and moves, and the gear revolution will guarantee enough collection moving intervals.So-called enough collection moving intervals are exactly that final position was wanted by the position of acquisition plane at the marginal point reflection visual angle of X-axis forward vector, promptly by the acquisition plane object at the marginal point of X-axis forward relatively and the position passed through about the reflection vector of this normal vector of the visual angle vector of camera.Camera parameter is set, it is focused, make camera carry out a data acquisition in per 5 seconds.

Data acquisition is as follows: linear light sorurce is moved on to acquisition plane X-axis negative sense one side, make projection line segment that the linear light sorurce reference position projects the Y=0 plane by edge one side of acquisition plane object on the X-axis negative sense, linear light sorurce is at the uniform velocity moved to the X-axis forward, gather a secondary data simultaneously every 5 seconds, arrive final position up to linear light sorurce, image data finishes.The displacement of the linear light source of general twice adjacent data is 1 millimeter, whole gatherer process about half an hour consuming time, gathers about 300 images.

(3) industrial camera is demarcated and data processing

In order to set up image coordinate system and the The Transformation Relation of Projection of gathering the space, the inside and outside ginseng that needs to determine industrial camera is demarcated, the present invention uses document 9-Zhengyou Zhang, the method of Gang Xu:A general expression of the fundamentalmatrix for both perspective and affine cameras.IJCAI 1997:pP.1502-1510 is gathered and is demarcated the used gridiron pattern data of the inside and outside ginseng of camera.Put into the black and white gridiron pattern and its center is overlapped with the center of gathering the space, gathers an image, mobile again gridiron pattern angle, many images of collection different angles are used these images and are recovered the confidential reference items of Daheng's camera and the camera outer ginseng with respect to the collection space.

For the data that collect in the step (2), carry out data processing by computing machine, be the data that collect are expressed as curve and storage with the form of pixel value track, its method is: a series of image is defined as Image _{U, v}(x), wherein x represents the current images data directory, supposes that the image number is n, and the scope of x is [1, n] so; U and v represent the position of current pixel point in image.

The geometric locus of each pixel comprises two peak values: first is the diffuse reflection peak value than broad and low peak value; second narrow and high peak value is the direct reflection peak value; the basis that two peak values that solve recover parameter as the BRDF modeling, x was x when the present invention established linear light sorurce and arrives directly over this pixel (being the diffuse reflection peak point) _d, x was x when linear light sorurce arrived minute surface transmit direction (being the direct reflection peak value) _m

Step 2: BRDF attribute modeling

The process of BRDF attribute modeling is as follows:

(1) sets up the Ward model

Carry out the attribute modeling for the data that collect, the isotropy ellipse Gauss sharpening function BRDF model that the present invention adopts Ward to propose.Because the Ward model is simply accurate, can recover the surface properties of real-world object preferably.The Ward model description the surperficial equation of going up the BRDF value of any point, promptly from reflection direction (θ _r, φ _r) outgoing luminous energy and from incident direction (θ _i, φ _i) the ratio f of luminous energy of incident _r(θ _i, φ _iθ _r, φ _r) as shown in the formula:

$f_r({\mathrm{\θ}}_i,{\mathrm{\φ}}_i;{\mathrm{\θ}}_r,{\mathrm{\φ}}_r)=\frac{{\mathrm{\ρ}}_d}{\mathrm{\π}}+{\mathrm{\ρ}}_s\·\frac{1}{\sqrt{\cos {\mathrm{\θ}}_i\cos {\mathrm{\θ}}_r}}\·\frac{\exp [{-\tan }^2\mathrm{\δ}/{\mathrm{\α}}^2]}{{4\mathrm{\π\α}}^2}$

In the following formula, ρ _dBe the diffuse reflection illumination parameter, ρ _sBe direct reflection illumination parameter, α is the expression material reflection coefficient of roughness, and δ is the angle of the half-angle vector h of normal vector n and incident ray and view directions.

(2) light source linearization and drafting reflection table

The light source that collecting device of the present invention uses is a linear light sorurce, and be that the light that the some directions to some sampled points launch calculates the value of the BRDF of corresponding a certain observation point in the Ward model, so need be understood as the integration summation of pointolite to linear light sorurce of the present invention.Carrying out integration respectively for the Lambert reflection model of diffuse reflection part and direct reflection part approaches.When Ward model of the present invention is carried out the integration summation of light source, if hypothesis viewpoint vector (θ _r, φ _r) and linear light sorurce be perpendicular, θ then _r=0, φ _rBe constant.Because the camera of image data is on the X=0 plane, therefore hypothesis is set up, and finally the computing formula of each pixel color value is as follows:

Image(x)＝ρ _dD(x)+ρ _sS _α(x)

The present invention is the BRDF integrated value after the light source integrationization is stored in the reflection table.Drawing as shown in Figure 3, the present invention of reflection table can pass through four steps:

(a) suppose diffuse reflection parameter ρ in the Ward model _d=1, partly carry out light source linearization integration for diffuse reflection, use pointolite is used as trace, the integration that whole linear light sorurce carries out in the overall region is found the solution, final result stores in the image, draws out diffuse reflection table D (x), shown in Fig. 3 left side D (x) part.

(b) suppose direct reflection parameter ρ in the Ward model _s=1, partly carry out light source linearization integration for direct reflection, carry out line integral for each x, the value of all of trying to achieve in the x field of definition stores in the image, draws out direct reflection table S _α(x), wherein the span of material coefficient of roughness α is (0,0.3).

(c) ask mathematical expectation μ and variances sigma for each x.

(d) above-mentioned two reflection tables are optimized, promptly draw empirical value through a large amount of experimental results, make it also can be well continuous when α ' approaches zero, as shown in the formula:

S _α' (x)=and 00＜α '＜0.005, α ' is the material coefficient of roughness after optimizing.

(3) diffuse reflection parameter ρ _d

The present invention at first obtains the diffuse reflection parameter ρ of Ward model _d, shown in Fig. 4-a.According to pixel geometric locus Image _{U, v}(x), can obtain the diffuse reflection peak value of each passage.Because collection of the present invention is the material of ideal flat, so, as can be known when linear light sorurce runs to certain some vertical direction, i.e. x=x _d, the pixel value that this point obtains is this diffuse reflection peak value, is made as L _dHere should be noted that linear light sorurce is put into sufficiently high position, make diffuse reflection peak value and direct reflection peak value separately enough far away, just can ignore of the influence of direct reflection peak value the diffuse reflection peak value.

In order to obtain ρ _dNeed also to know that (the present invention uses color calibration plate (ColorChecker), and to obtain known diffuse reflection parameter be ρ to pixel for u, what of luminous energy v) _StandardThe diffuse reflection peak L _Standard, obtain ρ according to proportional relation then _dValue as shown in the formula:

${\mathrm{\ρ}}_s={\mathrm{\ρ}}_{s\tan \mathrm{dard}}\·\frac{L_d}{L_{s\tan \mathrm{dard}}}$

(4) material coefficient of roughness α

Diffuse reflection parameter ρ _dAfter known, at first consider because of the diffuse reflection peak value with respect to the direct reflection peak width, so can not ignore of the influence of diffuse reflection peak value to the direct reflection peak value, in order to obtain accurate result, the present invention uses whole pixel value track and deducts the known diffuse reflection peak value of having tried to achieve, and obtains having only the geometric locus of direct reflection peak value.So the direct reflection geometric locus that deducts diffuse reflection part can be regarded as independent peak.

When linear light sorurce runs to direct reflection direction x _mDuring the place, the direct reflection peak value of this point produces, and the present invention obtains the mathematical expectation μ and the variances sigma of pixel value on this direct reflection geometric locus according to knowledge of statistics, and all pixel value sum S, tries to achieve according to following three formulas:

$\mathrm{\μ}=v\frac{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}x\·\mathrm{Image}\left(x\right)}{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)}$

${\mathrm{\σ}}^2=v\frac{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}{(x-\mathrm{\μ})}^2\·\mathrm{Image}\left(x\right)}{\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)}$

$S=v\underset{x={2x}_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)$

X wherein _mBe the position at direct reflection peak value place, get with respect to x _mPixel value in the zone of symmetry is promptly from x _m=2x _m-x _dTo x _d, because industrial camera is to the influence of this regional projective transformation, the width that different pixels is corresponding different, so all need scaledly for all values, zoom factor is a following formula:

$v=\frac{1}{2(x_d-x_m)}$

After obtaining these statistical sample data, from the reflection table of the linear light sorurce drawn out before, table look-up and obtain σ in the corresponding reflection table of the σ that calculates with statistical sample, be one to one because of σ in the linear light sorurce reflection table and α again, so just can obtain α, and unique, shown in Fig. 4-c.

(5) direct reflection parameter ρ _s

Direct reflection parameter ρ _sThe method that obtains with diffuse reflection parameter ρ _dMethod is similar, shown in Fig. 4-b.But at first to consider to obtain direct reflection parameter ρ corresponding in the corresponding linear light sorurce reflection table with known α _s, in order to find ρ _sNeed counting statistics sample summation S, search the reflection table then, find out the S of wherein identical α correspondence _α(x), use and find the solution ρ at last _dSame method is tried to achieve ρ _s, as shown in the formula:

${\mathrm{\ρ}}_s={\mathrm{\ρ}}_{s\tan \mathrm{dard}}\·\frac{\mathrm{Ls}}{L_{s\tan \mathrm{dard}}}$

Step 3: GPU real-time rendering

Draw according to Ward model modeling result, the parameter of recovering is stored with the texture form, the Fragmentshader stage at GPU loads data texturing, and pixel color value calculated in real time, the experimental result such as Fig. 5-b that draw, the real pictures of being gathered object of Fig. 5-a for taking, the lower part of true picture of having drawn the reproduction true to nature of the lower part of image screenshotss of experimental result.

Claims (5)

1, a kind of true object material quality reflecting attribute modeling approach based on linear light sorurce is characterized in that step is as follows:

(1) based on the data acquisition equipment of linear light sorurce to data acquisition and processing (DAP)

The data acquisition equipment of described linear light sorurce comprises:

The Le Gao robot is as the control system and the automatic translation system of data acquisition equipment;

Industrial camera is as the view data capture systems;

Computing machine is as the data handling system of data acquisition equipment;

Linear light sorurce is as the light-source system of data acquisition equipment;

Linear light sorurce sends linear light irradiation on the quilt planar object of gathering, control system drives automatic translation system and moves forward, industrial camera carries out image acquisition to the plane of being gathered simultaneously, and computing machine carries out follow-up data to the view data of gathering to be handled, and obtains pixel value geometric locus Image _{U, v}(x) also storage, wherein x represents the current images data directory, and the scope of x is [1, n], and n is the image number, and u and v represent the position of current pixel point in image;

(2) utilize linear light sorurce in the step (1), generate the reflection table;

(3) BRDF attribute modeling;

According to pixel value geometric locus and the middle reflection table that generates of step (2) that step (1) obtains, obtain the ρ of Ward illumination model _d, and then obtain α, obtain ρ at last _s, ρ wherein _dThe diffuse reflection parameter, ρ _sThe direct reflection parameter, α is the material coefficient of roughness.

2, the true object material quality reflecting attribute modeling approach based on linear light sorurce according to claim 1 is characterized in that the method that generates the reflection table in the described step (2) is:

(a) suppose diffuse reflection parameter ρ in the Ward model _d=1, partly utilize linear light sorurce to carry out light source linearization integration for diffuse reflection, draw out diffuse reflection table D (x);

(b) suppose direct reflection parameter ρ in the Ward model _s=1, partly utilize linear light sorurce to carry out light source linearization integration for direct reflection, draw out direct reflection table S _α(x), wherein α is the material coefficient of roughness, and its span is 0-0.3;

(c) for each current images data directory x, ask mathematical expectation μ and variances sigma;

(d) to step (a) and two reflection tables that generate (b) be optimized, it is as follows to optimize formula:

S _{α '}(x)=and 00＜α '＜0.005, α ' is the material coefficient of roughness after optimizing.

3, the true object material quality reflecting attribute modeling approach based on linear light sorurce according to claim 1 is characterized in that described step (3) diffuse reflection parameter ρ _dObtain as follows:

ρ _dValue as shown in the formula:

${\mathrm{\ρ}}_d={\mathrm{\ρ}}_{s\tan \mathrm{dard}}\·\frac{L_d}{L_{s\tan \mathrm{dard}}}$

ρ _StandardBe known diffuse reflection parameter, L _StandardBe known diffuse reflection parameter ρ _StandardReflection peak, L _dThe diffuse reflection peak value of the pixel value geometric locus that obtains after handling for computer data.

4, the true object material quality reflecting attribute modeling approach based on linear light sorurce according to claim 1 is characterized in that the method that obtains of the material coefficient of roughness α of described step (3) is: when linear light sorurce runs to direct reflection direction x _mDuring the place, the direct reflection peak value of this point produces, and according to knowledge of statistics, obtains the mathematical expectation μ and the variances sigma of pixel value on this direct reflection geometric locus, and all pixel value sum S, promptly tries to achieve according to following three formulas:

$\mathrm{\μ}=v\frac{\underset{x=2x_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}x\·\mathrm{Image}\left(x\right)}{\underset{x=2x_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)}$

${\mathrm{\σ}}^2=v\frac{\underset{x=2x_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}{(x-\mathrm{\μ})}^2\·\mathrm{Image}\left(x\right)}{\underset{x=2x_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)}$

$S=v\underset{x=2x_m-x_d}{\overset{x_d}{\mathrm{\Σ}}}\mathrm{Image}\left(x\right)$

X wherein _mBe the position at direct reflection peak value place, x _dBe diffuse reflection peak value position, Image (x) is the x width of cloth view data of gathering, and gets with respect to x _mPixel value in the zone of symmetry is promptly from x _m=2x _m-x _dTo x _d, because camera is to the influence of this regional projective transformation, the width that different pixels is corresponding different, so all need scaledly for all values, zoom factor is a following formula:

$v=\frac{1}{2(x_d-x_m)}$

After obtaining these statistical sample data, from the reflection table of the linear light sorurce drawn out before, the σ in the corresponding reflection table of σ that obtains calculating with statistical sample of tabling look-up is again because σ and α are one to one in the linear light sorurce reflection table, so just can obtain α, and unique.

5, the true object material quality reflecting attribute modeling approach based on linear light sorurce according to claim 1 is characterized in that described step (3) direct reflection parameter ρ _sFormula:

${\mathrm{\ρ}}_s={\mathrm{\ρ}\′}_{s\tan \mathrm{dard}}\·\frac{\mathrm{Ls}}{{L\′}_{s\tan \mathrm{dard}}}$

Wherein:

Be known minute surface diffuse reflection parameter,

Be known minute surface diffuse reflection parameter Reflection peak, Ls is the direct reflection peak value of the pixel value geometric locus that obtains after computer data is handled.

Images