TWI315042B - Method of three-dimensional digital human model construction from two photos and obtaining anthropometry information - Google Patents

Description

1315042 IX. Description of invention: [Technical field to which the invention belongs], 目, 丨 if 明 关于 关于 关于 关于 关于 if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if The feature points, feature height and feature width make the three surnames, while τΐϊί saves the complete two-dimensional (10) features, and produces a reasonable three-dimensional digital human body model; and roots rm] purely long-aged, Lin Three-dimensional human test results. Dimensions, !! 'Where the development of all kinds of clothing design or the number of defense industry needs, according to 4 =, the establishment of the human body of the human body; w in addition to the 'mouth mouth and mouth juice : Also have 1 = 22: = measure the phase _ ί = fee = loss as r improve === lack of only ====== side take feature points are still missing manually selected by the same person It is difficult to repeatedly select the same point in the million cloud 4 1315042 points, not to mention the different people's perception of the same information, and its heterogeneity is absolutely greater. Therefore, the domestic researcher Chen Zhiju (see Attachment Reference ^ 31) proposed a method of automatically searching for features on the cloud point of the human body and structuring the human body according to the characteristics. This method improved the lack of the above manual selection. There are also Caesar (CAES AR) programs that have been implemented in the United States and the European Union for several years (see Attachment Reference 33). 'Pipe on the scanning of human features 72 points to reduce the time and effort required for manual selection = work, but it is still derived The same problem as manual selection, that is, the marked points can be misplaced, may be missed, and the implementation personnel's perception of human features is also poor.

It is still not ideal. In addition, since the human body scanner is not a necessity for people's livelihood, its price is mutex and not popular. So it is difficult to adopt a structural change by changing the three-dimensional digital body by scanning cloud points. Digital human technology has covered all possible feature searches, and its structural robustness still needs to be strengthened. However, it is not pourable. In recent years, the technique of three-dimensional human body surface has many aspects. According to the attached reference 19, the technology for reconstructing the three-dimensional human body model is divided into = main directions. - After obtaining the scanning point data for the marriage of the three-dimensional person, the learning method reconstructs the human body view. The second ride to the human body scanner is not enough, and in order to build a digital body more quickly, and in the reconstruction method ίίΖίϋί industry knowledge, in order to use the photo to replace the scanner as a reconstruction of the human distribution, in order to develop the _ insert if it ^ To reconstruct three-dimensional newcomers __ research, = technology can be divided into two kinds of - is the job cloud point to go to the noise, feature search eight surface fitting and other steps, directly based on the scanning point data is based on The secret tree paving type, with the most = ^ n

Wa^g (see Attachment Reference 26) #人 In the reconstruction of the human body model, the method of J is divided into several steps to reconstruct 'including: removing the noise; sweeping the positive, making the direction to facilitate the subsequent feature search, · segmentation The point group is the head, the lin, the f hand, the right hand 1315042, the left leg, the right leg, six blocks. Six block grids are created by using the method of selecting the points of the cylindrical coordinates; the s-block data points are connected at the s-dump block, and the blocks are added, and finally the areas missing from the data on the scan points are complemented. When removing the segmentation point group, 'the concept of Saki is applied, and the related member letter, number, and (membership fimCti〇n) are used to help the judgment. Wang et al. also developed a method of re-constructing the three-dimensional human body model. (See Attachment Reference 27) 'There are three steps in reconstructing the human body model: (the concept of one is to find the feature points from the scanning point data and construct the person using the feature points; (2) to connect the feature points on the original scanning cloud point Line to determine the characteristic curve, use the characteristic curve to modify the topology to get the initial human body model〗

Compare the initial anatomy of the human body with the scaled data of the optimization method, and adjust the shape of the curved surface to obtain the details. The most common problem between the scanning point and the establishment of the parameter D model is that the created meshes will intersect each other and cause the next _face fitting to produce erroneous results, even failing to correctly formulate two. So Barhak Etc. (see Attachment 3) proposes two ways to establish 夂, 格, to avoid this problem, to use a partial micro, and to use a neural network (neuralnetw〇rk) To build. When the human body scans, it is usually impossible to obtain a complete scan in areas such as underarms and underarms. It is the most common problem encountered in reconstructing three-dimensional human bodies from scan points. ° type to fit the close charm, is also the sharp three miscellaneous people _ jug two j as Seo et al. (see Attachment Reference 21) proposed model model, reconstruct 3D human ugly method, the structure of its model Divided into skeleton and skin ^ soil grid, the moon frame is used to determine the posture of the three-dimensional human body, and the skin mesh is used to determine the shape of the body surface. Their fine optimization method maps the skeleton feature points of the model model to the scanning points. The location of the feature points, and the skin mesh of the template model is mapped to the scan point data by an optimization method, and the details are modified. And this construction method ^J You· is that 'the relative relationship between the feature points between the skeleton and the skin mesh can be defined in advance and the process is fully automatic'. Therefore, as long as the parameters of the bell are modified, it is difficult to repair the entire construction. . Another: u et al. (see Attachment Reference 12) proposed 1315042 is a heavy, two-dimensional human body is the use of data point global mapping (globalmapping) local deformation (1〇caldef〇rmati〇n) technology. The overall mapping method is that the scan = data has no extinction points d (four) should be _, and the corresponding relationship is ^ vector to indicate the magnitude of the displacement of the feature points in the X, γ, Z directions. For the amount of deformation outside the feature point, the radial basis function (radialbasis ^tK>, η) is used to establish the interpolation method. After the initial deformation, the sample feature points can be automatically moved to the same position as the feature points of the scanned data, and the details are modified by local deformation. In terms of local deformation, the node (nQde) of the mesh on the fine surface is searched closest to the point data, and then it is completed. There is also Allen et al. (see attachment ^ text =) _ l by optimizing the square wire to fit the grid of the data board, and sweep: point markers (markers) such a method for sweep = = Post-processing is very convenient, which simplifies the time when feature point search and complement points are processed. Attachment Reference 1 uses the optimized domain description point location, the objective function is composed of three, and combinations; contains the data point error (dat read. Small smoothing error is still ^) t ^ ticket Point error (marker_r). And the effect of each item in the fit. Adjust different weights for different parts to achieve a better type. The advantages of the body are two. In addition to the information obtained by using the model model, you can also get the color in the color photo = lto=rr= More about the left and right? (see Attachment 11). By modifying the model model by Φ /ra-r '« 像 像 数 数 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改 修改In the jade, the 刭 刭 雒 雒 && 313 human body model. The method is to first project the relative relationship of the model image and the three-dimensional model model to the second version of the image deformation. Because the relationship from ya is known, it can be reversed from the deformed 7 1315042 projection image back to the three-dimensional shape. Finally, add the color from the upper phase to complete a simple 3D model that can be used to create animations. In addition, the method of constructing the human age from the B (see Annex U) is obtained by five (head and side _ Zhang 'body front, side and back three); ^, provide users manually Select feature points on the photo to find the feature outline in the photo. The reference person's miscellaneous profile is used to make the pattern of the cut-off of the pattern and the color map to obtain the final result. As for Wang et al. (see the attached reference text t for the front and side photos to obtain the feature points of the re-modification model model to achieve the purpose of reconstructing the 3〇 mannequin. Since the feature line has been defined on the model model, it is reconstructed The _ type can be used very well in the design of clothing. For the technique of modifying the 3D model modal 3L according to the two-dimensional information, since the known information has only two orthogonal planes, the 丄 coordinate conversion _ three tonic is used. Wei miscellaneous, but the difference between different feet f is non-hetero, and the method will produce large errors in some cases. The method of reconstructing the three-dimensional human body model based on statistical analysis is by mixing ( Blendmg) Two or more human models to generate the target human body model. Therefore, this method needs to collect a large number of digital gambling types, and analyze the characteristics of the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The characteristics of the Nigga miscellaneous supply, the selection of similar human body models in the value domain for mixing. Therefore, ^agnement-Thalmann et al. (see Attachment Reference 18) utilizes the unified $Dt bribery method, and The virtual test of the actual wire is turned on: the result of the user reconstructing the approximate figure by adjusting eight parameters. On reconstructing the three-dimensional human body model, reference 18 makes the method a two-step process. To complete, firstly, the template should be molded to the scanning point, and the parameters of the three people should be the same; the final analysis of the statistical information on the human intelligence, respectively, should be / a parameter to the phase riding conversion matrix 'you can get the individual crime model 2, Although the human body model obtained by the method will be in a reasonable range, it does not surely show the individual's turning characteristics. Therefore, in the financial screaming, it produces a considerable degree of error in 8 1315042. There are also Seo et al. Attachment References 21, 22, 23) Using anthropometry to obtain the characteristics of the 3D human body, and then performing a principal component analysis (PCA) to obtain a representative set of feature vectors. The interface is to change the feature vector. In order to adjust the appearance of the 3D human body model, the internal calculation is to build a feature space from the database of 3D human body models with various heights, and the amount of internal database. It directly affects the accuracy of the final result. The eigenvector input by the user will fall at a certain point in the space after analysis, and finally the interpolation is performed by several known human models in the database. The results of these references, in addition to the use of this particular pose space deformation (p〇sespacedef〇rmati〇n; PSD) in more representative human body feature lengths, girth, etc. Sexual parameters are introduced into this deformation method, including fatpercentage and waist (hip_waist_mtio). These two parameters and PSD can simulate a person's fat change on a computer. The results of free-form surface deformation are often used to simulate changes in human motion and expression. In the dynamics of human body surface, the methods used in these researches have a common shortcoming, that is, in some places where the surface is curved, there will be unreasonable ^ sag i Lewis et al. (see Attachment 15). The concept, ^ from the rich known of the > material to create a posture space printing coffee), any kind of 3 space (10) ride can be expected to know some of the known information, interpolated, can be used Shepard's method, radial basis notion ' ^1:^^ (energyfunctions) (convex :methods) 'so the result will be more realistic. In the part of the human body, a more realistic 3D human body model can be built from the known feature data by establishing a similar feature space and then inserting it through appropriate axis insertion. &+ί by BlanZ et al. (see Attachment Reference 5) for the reconstruction of facial models. Conceptually, we first establish a database of ίτ Π model, and then analyze the babies of the Detective Department, and then build a face that can be constructed by the 4' system. The (3) alias is generated by the 槪 vector _ group 3D face sample 1315042 ^. The feature vector of the model in the face is divided into two parts: two-dimensional map and three-dimensional shape = ' and both of the wealth - corresponding _ system. Therefore, in addition to the user manually adjusting the parameters to construct the three-face view, the two-dimensional texture of the template model can be changed by inputting the _ face photo, similar to the image morphing technique, and then by The relationship between the 2D texture and the 3D geometry yields the desired facial model. The various methods for reconstructing the three-dimensional digital human body model disclosed in many of the above-mentioned conventional references are in the display of a product that is of great industrial value and worthy of betting. Therefore, the present invention also improves and develops how to construct a three-dimensional digital human body model by using at least two photos f; and further obtains a three-dimensional human body measurement result according to the minimum outsourcing characteristic length measurement formula established on the model, and hopes to do To the connection and realization of three-dimensional human measurement technology with two photos. In response to this, efforts have been made in many aspects to achieve the results of the present invention. SUMMARY OF THE INVENTION The main purpose of the present invention is to provide a method for constructing a three-dimensional digital model from a photograph, mainly to (4) quickly and efficiently reconstruct a two-dimensional digital human body model of approximate photos, and greatly enhance the human body's subtle immersiveness. A further object of the present invention is to establish a two-dimensional human body measurement for each feature of the human body and to develop a related mathematical formula, which can be used to search for the contour of the side image, the height and the feature width, so that the reconstructed three-dimensional image The digital human body model completely preserves the two-dimensional contour features' and produces reasonable three-dimensional results to: the degree of emulation of the dimensional shape at the subtle detail. - Dimensional Model Another object of the present invention is to derive a mathematical equation for the amount of deformation, to quickly complete the deformation of the model model, and the three reduction positions of the charm student. According to still another object of the present invention, the three-dimensional human body measurement result is further obtained according to the minimum feature phase formula established on the model, so as to achieve the connection and realization of the three-dimensional human body measurement technology by the two-dimensional generation, thereby establishing Three-dimensional human body measurement & One of the inventions is a method for constructing a three-dimensional digital human body model from a photo, the main image obtained by the human body positive film of the main 1315042 and the side photo 丄), which is obtained by the circle profile: the above-mentioned special width Search f \ from the contour to obtain feature points, feature height, whole model, and reconstructed two

The project specializes in defining and transferring a number of ship points, heights and feature widths. The definition of the test equation is sharp. The definition of the measurement items and the number of calculations are as follows: The amount of projection obtained from the front: (1) Shantou vertices (HeadPoint)·· The contour of the human body, the point of maximum devaluation, the same Ζ value is not only - when the point 'takes the average of all points as the head apex; ' ^1.0 eQ /, ^ i. 〇 = {Piiy.z) I Vz, Z = (Ω^-)}

If Num(Fl0)^l , then y(Fl0)

b) HeadBreadth: Above the head and above the ear, the maximum horizontal line distance, that is, from the height of the head apex, searches for the position where the first horizontal width and the next layer change rate are negative, and the width is recorded as the head width. ; ^=Muc(z(/)) (b) Fzol F2.0R (left and right) Armpit Point: the intersection of the left and right inner arm curves and the torso', ie the left hem point group from the minimum γ value Point to the point group between the minimum Z values, find the maximum ζ worth the left 腋 point f2.gl, the right half point group from the point of the maximum Y value to the point group between the minimum z value to find the maximum ζ 11 1315042 worth the right腋下点 f2.0R;

Suppose Ω^1ϊ_£={/>|/>€Ω^> = ^ = => Z{F2.0l) = ^max (P-armpit_L )

Suppose Q^,_s={i>|/>e^} «<ί<^α = ^(^(Ω/κ))>Α=:ι·„^(ΖίΒίη(Ω^^ Z ( -^2.0« ) = Zmax (^ampit_R ) (4) f2 neck width _1^_: between the head width Fl height to the left shoulder point ^ height between the height of the human body, the right and left body contours are the shortest Lu distance; fine _ = > 邶) > paper. i)}

=>F2=Min(HdisU • (5)F3.〇L F3.0R (left and right) Shoulder points (Sh〇ulder p〇int): Left and right shoulders ^ The point with the greatest curvature, ie the left half point In the group, the point group whose hdex value is between the head vertex and the minimum Y value, respectively find the point with the same value of the left side point Fn and the point of the same Z value, and find between the two points The BV value is the highest at h.oL. In the right half-body group, the point group between the point where the index value is between the head apex and the maximum γ value is found separately from the point of the right squat point & And the point of the same Z value 'and find the BV value between the two points to get the maximum F3.or;

Suppose Q^_iWrfer = 1 ^ 6 « < ^ < A 1 [α < index(Fz〇L), y(a) = y(F2S)l),fi< index(F2,L),z( Fi) = z(F2.0L)j ^F3.〇l ~Max[BV(P^

Suppose ςΐ·α<ί<β' [«>z^ex(F2,,)>z(a) = z(F,0,)^>zWex(F2,J!),j;(^ ) = 7(F2,,) =>F30R =Max[BV(P^) (six) F3 ShoulderBreadth: The straight line distance between the ends of the left and right shoulders, that is, the left and right sides of the human contour point are calculated first. Shoulder point f3〇lF3〇r, and find the linear distance between 12 1315042; F3 = dist(F3〇L, F3〇r) (7) F4 chest width (BustBreadth): between left and right The straight line distance, that is, first calculate the left and right points F2.〇LF2.〇R on the contour points of the human body, and find the straight line distance between them; F4=dist(FZ0L, F2QR)

(8) Fs.o CrotchPoint: The intersection with the highest curve in the left and right legs, that is, the point where the Z value of the left and right inner leg curves is the largest, from the point to the -Z direction. 30 height range, seeking the middle point of all sorting points of the curve in the range of the leg as the underarm point; defining the curve in the left and right legs as: Under the point contour sorting condition, from the lowest point of the right half to the lowest point of the left half of the body The curve formed by the point group.

Suppose Qlnside Jegli„e = ^ = Max{IWex(zmin(n/R)), IWex(zinin(Q/t))} I /? € ^insidejegUne9CC <1<β9

Suppose ^crotch_point=-a<index{z^{aim^^^ .

β > index{ZmxK {^_^)),Ηάι5ί{β,ζ^ {aimideJegline)) = •5U ^ ^^^n^eX{^,crotch_poml)

(9) & Horizontal Waist Breadth: The shortest horizontal straight line distance between the left and right sides of the torso between the underarm and the underarm height, that is, the Z value of the side waist width Si corresponds to the front contour Waist width; P0 e Ω^, e Ω^, z(Pa) = z(P6) = z(St) => FS= Min{Hdistaf (Pa, Pb)) 13 1315042 (10) F6 natural waist width (Natural Waist Breadth): The shortest straight line distance between the left and right sides of the torso between the underarm and the middle of the point, that is, the Z value at the point F2.0]L F2.or to the underarm point F5.〇 Find the shortest distance between the contour points on the left and right sides of the human torso between heights;

Supposea coffee -={(Ρ(\Ρ々Ω/,υ:ζ(^) >z(Pi)>z{FSSi))) Μ => F6 = Mm{dist(Qnearwaistline)) (H ) F7 Hip Breadth: Below the horizontal waist width, above the kneeling point, the maximum horizontal line distance between the left and right sides of the trunk, that is, the Z value of the side hip width S4 corresponds to the hip width of the front contour;

Suppose y(,Pa)^y{.Pb), y(Pc) = y(Pcenimid), z(Pa) = z(Pb) = z(Pc) = z(S4) A e Ω where, g Ω negative , /^ e Ω, F7 = Min(Hdista/ (Pa, Pc)) + Min(HdistQf (Pb, Pc)) (12) FSL FSR (left and right) Thigh Breadth: · Underarm Point to the left (right, right) horizontal distance between the left and right sides of the thigh, that is, the horizontal width of the left and right legs at the same height as the underarm F50;

Suppose 〇near J^high = IΧ/?) ^ y(FS0), Z(P:) = z(F50), /> e Ω^} => Fsl =Min(Hdista/(Fs〇fPa)) , Pa ^nettrMh

Suppose ^near rightthigh = {Pt I y(Pt) ^y(F5〇)y Z(J>) = z(F50),i> € Ω^} => FSR = Min(Hdista/ (FS0,Pb) ), Pb e 〇.near rightth.gh (13) Fpr (left, right) Low Knee Breadth: The horizontal spacing at the bottom of the knees of the legs, that is, by the lower point F5 〇 along the left and right Find the maximum BV value of the inside of the leg between 1/8 and 3/8 height and the contour curve opening in the +Y direction (Fpl) and the Y-direction (f9r), and find the left and right knee levels respectively. Wide distance

Suppose Q,.BjWe Jeftlegline = {Pt | Pt s ^imide_legUne I , 1315042 (My 2 legs (Ω/ )+! WW) < z(巧) < ( 眶z眶(Ω, )+. ζ- ( Ω,))}

Let k = 35^P^a inside _ WMegHne BP^BViP^BVi^) F9L = Hdista/ (Pa, Pb), Pb e Clβ where y(Pa) Φ y{Pb\ z(Pa) = z{Pb) , Ptt, Pbe Ωβ

Suppose 〇·inside_n^i\le^me ~ {^i I e ^inside Jegline ^ » (^TMχ(Ω^+^πώ(Ω/)<ζ(/>)<(|ζπ1ίχ(Ω /)+|ζπώ1(Ω/))} • Let ^ = 35 => VP, e Ω^β rightfe?/ftie 3 Pc a BV(PC) >BV(P{) F9R ^ Hdista^ Pd • ^ Here y(Pc) ^ y(Pd), z(Pc) = z(Pd), Pc9Pde 4 (fourteen) F10LF10R (left and right) Mid Thigh Breadth: 胯 down to (left,

Right) The horizontal line distance between the midpoints of the knee width is the lower point f5.〇 and the lower knee width f15L

The Z value of the midpoint of the height between the Fims is taken as the horizontal width of the left and right thighs, respectively;

Fl0L =Hdist(Pa,Pb) where y(Pa)^y(Pb), z(Pa) = z(Pb) = } Ρα,Ρ,〇^ φ F10, =Hdist(Pc,Pd) where y(Pc )^y(Pd), z(Pc) = z(Prf) = ^)^z(F^), Pc, (fifteen) Fii_〇l Fii.qr (left and right) heel point : In the left and right calf to sole curve, the point of maximum curvature, that is, about 1/8 body length from the lowest point of the left and right half of the body, in this range, the maximum Bv value in the left and right legs is obtained;

Suppose Q„earJeftheel = {/>Ii>eQinift/e_We,z(i>)<(^ηΒΧ(Ω/)+Ιζπ.η(Ω/))}

Let k = 35 two outside (5) (10) brown coffee 3 FuQL = Max (Bvm 15 1315042

Suppose Clnear_Hghthed = {i> I /> e < (^ΙΜχ(Ω/)+|ζΠ]ίη(Ω/))}

Le“ = 35 3 510/i=Mu(5i^)) (16) F11LF11R (left and right) CalfBreadth: The maximum horizontal distance between the left and right sides of the calf is 'left and right low knee width F9L F9R See you on the heel

Fii.or's Z value height, respectively, the maximum horizontal width of the left and right calves;

SupposeΩ_ w = {(Pt I p{ € Ωα,ζ(^£) > z(^.) > z(Fn.0i))} Ful = Mox(Hdist(nnearJeftcaI/))

Suppose 〇.near rishtcalf = {(i> IP. e Ω^,ζ(^Λ) >> z(Fn 0Jt))}

Fnji =Max(Hdist(^nearrlghtca!f)) (17) F12lF12R (left and right) Min Calf Breadth: The minimum horizontal distance between the left and right sides of the calf, that is, the left and right knee width? 11. 〇 1 ^ „. The Z value of the service, the minimum horizontal distance between the left and right calves;

Suppose ^near_1φηαηcalf = {(^/ I , ζ(Ρ9ί ) > z(/^·) > 〇i ))} FUL = Μίη(Η—(Ω_邮一lf)

Suppose ClneBrrightltiacalf = {(/> I z{F9R) > z(^.) > z(F110/!))}

FllR = Min(Hdist(Cl酣弟-!f)) (18)? 13. 〇1^13 dish (left, right) side neck point (^〇1^0 him): the most curved point of the left and right neck, that is, the height of the neck width F2 to the shoulder point F3.〇LF3. Between the heights of R, find the point where the BV value of k=20, k=35, and k=50 is the largest, and take the average of the three points;

Suppose QnearJ^,eck = lii I e < z(Pt) < (^ζηΒΧ(Ω/) + ^ζηιίη(Ω/))1 1315042

3^.0, U 象(万·U ' 3 Ω"—一十|/^Ω_(~Λ)<2β)<(^_(Ω/)+^2ηώι(Ω/))} =&gt ;Vi>eQ„eflrr;gAftierf Ξ PlX0R=幽(4)))丨U—+Ma(·))丨(4) 3 (19)Fu torso length (TnmkLength): from the average side neck height to the vertical point of the underarm High skin difference, that is, the Z value difference between the average height of the left and right side necks SF13.〇lF13.〇r to the underarm point p5();

Fi3 = z(pw^(f·)—bright.)

(20) Hip Height: The difference from the hip width to the vertical height of the heel, that is, the hip width FV level to the heel point; the mean value of the mean height of Fu.olFh.or; & = along W) +汾W) _峨) (2Ί1)1?151^1511 (left and right) knee width 〇^1^ ugly ratio 3(^11): at the height of the humerus, the horizontal distance between the left and right legs, That is, from the lower point F5_〇, find the length of 1/8 to 3/8 height along the inner side of the left and right legs, and the contour curve opening is respectively in the +Y direction (Ful), and in the -Y direction (F^r is obtained) The maximum BV value, respectively, the horizontal width of the left and right knees;

Suppose nillsideJeJUegline I e ^insideJegiine I β, (g Zmax (Ω/) + g ^min (Ω/) < ^(^) < (~ (Ω/) +1(Q^))

Let k = 35 two M e Ω (four) em coffee 3 Pa 3 BV (Pa) k BV (P () ^isL ~ Hdist^ {Pa9Pb\

Wherey{Pa)^y(Pb),z(Pa) = z(pb),Pa9Pi>eQ/L

Suppose 〇.inside_rigMegUne = Ρί \ Pi Dad Qinside Jegline I Q β, (g Zmax (Ω/) + -^nrin (〇/) < ^i) < (~ ^max (Ω/) +1(Qr))

Let Λ = 35 =>V/> e Qinside_righllegline 3 Pc a BV{PC) >5F(/>) 17 1315042 ^ΐ5Λ = Hdist^(Pc9Pd)9 Pd s where y(Pc) Φ y{Pd \z{Pc) = z{Pd\Pc9 Pd € (22) F16LF16R (left and right) KneeHeight: The difference in height from the knee width to the heel, ie the left and right knee width F15LF15R level to The Z value difference between the left and right heel points Fii.olFii.or; ^ Ζ(^1Ι.〇ΐ)~Ζ(·^15ΐ)

Fl6R — z(Fuor)~~ z(F15r) (23) Fn Scye Depth: The vertical height from the side neck point to the underarm point, the difference between the left and right neck points FmFn.oR Connection, left and right armpits 'point F2.0LF2.0R connection, take the z value difference between the two points of the midpoint; F17 = Vdht{-nsif· ^ F^r , Ful + ^2-0Jt) (two 14) F1S thigh length (ThighLength): the vertical height difference from the kneeling point to the knee width, that is, the Z value difference between the kneeling point F5. 〇 to the knee width F15L F15R average height; /γ8 = ζ(ρ5〇)_ £(^)±<5μ) (25) F19胯High (inner leg length) (CotchHeight): The vertical height difference from the underarm point to the heel point, that is, the lower point F5. 〇 to the left and right heels Point fuol

The difference in the Z value of the average height of Fii.or;

Fl9 — l.ot 9 Fs〇) + Vdist{Fn〇R 9 Fso) (26) F2 ShoulderHeight: The average of the two vertical height differences from the shoulder to the heel, ie the shoulders Point FmlF3. The average height of the sigh to the Z value of the average height of the two heels Fu.olFii.or; ρ _^^ί(β[ΙΛ1ι,υ + ν(ϋ3ί(βηΛΛ,υ 20 2 (27) FZ1 neck NeckHeight: The difference in vertical height from the side neck to the heel point', ie the average height of the Fol Fror on both sides to the Z value of the average height of the two heels Fii.olFii.or; F2i _ Qiao 3.0!) + version Pair 18 1315042 (28) F22 B 距 (BodyRise): from the waist width level to the underside point of the vertical difference, that is, the horizontal waist width Fs level height to the subordinate point F5 〇 Z value gap; F22 =z(F5)~z(F5〇) Second, the amount of projection on the side (1) SL head vertex (side Head Point) 1. The contour of the human body, the maximum value of z, if the same Z value is not only a little , taking the middle of all points as the head vertex; su〇e Ω/} 5,0 = {/>(χ,ζ) I Vz, z =

(2) 31 side waist wide position (16\¥ & 1313 than & her): between the lower point of the armpit and the height of the underarm height, the shortest horizontal spacing between the front and back of the side trunk, that is, the lower point Fzol Fzor Between the average south degree and the lower point F^o, the z-value of the posterior body contour curve '丨 丨 between 1/2 and 3/4', if face right(), take the coordinate of the minimum X value Point 'if not face right(), take the coordinate point of the maximum X value, and the horizontal distance from the curve of the precursor;

If face_right〇, x(Pa) = Μαχ(^.)), Pt 6 Ω„ββΓ if not face_right〇, x(Pa) = Mn(^.)), P-, € ^ 5, = Hdist^ (pa , ), where z(Pb) = z(Pe), Pb e Ω5/ (3) S2. Side Heel Point · · In the calf to sole curve, the point of maximum curvature, that is, the minimum Z value of the contour To the bottom 1/8 range of the maximum Z-value of the contour, find the maximum BV value of the back-body curve; 19 1315042 %?pose Ω ifif_=still|6,ζ—(Ω,)<z(6)<(我2 Bribe (ω^) + ΖπΖι(Ω:))}

Let k = 35 => M e Qnear_sldeheei 3 S2_0 = (4) S2 height (HumanHeight): the height difference between the heel point and the head apex, ie the distance from the heel S2.0 to the top of the head Si.o at the Z value; S2 = ζ {^ιΛ) — ζ(β2Λ) (5) S3 Waist Height: The difference from the lateral waist width to the vertical height of the heel, that is, the horizontal waist width S! height to the heel point S2. 〇 Z value gap; = z (5,)-z(5i0) (6) S4 Side Hip Breadth: Below the horizontal waist width, above the neck, the longest horizontal straight line distance between the curve of the front body and the curve of the back body' At the side waist width S! to the point group of the Z value of the lower point F5.0, if face right (), take the coordinate point of the maximum X value, if not face rightO 'take the coordinate point of the smallest X value' The horizontal distance from the curve of the precursor;

Qnear_sutehip = I ^ « Ωι4, ζ(5,) >ζ(/>)> Z(F5〇)} if face_right〇, x(Pa) ^Max(x(I>)), P( if not Face_right〇, x(Pa) = Μΐη(χ〇>)), /> ^near_sidehip S^Hdista {Pa,Pb), where z(PA) = z(PJ, Pb^Qsf (seven) 85 leg width (3 丨 (16 11% 11 ugly ratio & (1 &): at the height of the squat point, the horizontal width of the thighs, that is, the two rounds of the height of the Fa with the Z value relative to the front squat point, calculate Its horizontal width;

Suppose {P^lxiPJ^x^ z(Pa) = z(Pb) = z(F5M), Pa^Qsf, Pbensb} ^S5=HdiStns(Pa,Pb) (8) Se Knee Breadth: The horizontal width of the side of the most prominent point of the knee, that is, the two contour points of the same z value of 20 1315042 with respect to the average height of the front knee width Ful'Fur, and the horizontal width is calculated;

Suppose {^,/»1^)^^), z(Pa) = z(Pb)=z{F^)+2Z{FiSX\pa eQsf,Pb eΩϊ6| ^S^Hdistas(Pa,Pb) (nine S7 Middle Wide (SideMidThighBreadth): the horizontal width of the midpoint of the height between the lower jaw and the lateral knee width, that is, the two contour points with the height of the leg width Fl〇1 and the Z value in the front face, and the horizontal width is calculated; S7=Hdista(Pa,Pi)) where ^Pa) ^ x(Pb), z(Pa) = z(Pb) = z(Fl0L), Pa ^Cisf, Pb eQJft (ten) S8 minimum leg width (SideMinCalfBreadth) : the minimum horizontal width between the knee width and the lowest point of the contour, that is, the minimum horizontal distance between the width of the leg and the z value of the heel S2Q;

Suppose Ω£ο// = {P. I p. e Qsf,z(S20) < z(^) < z(55)} 58 =Mn(brain, (Ωω//)) (H^一) Side Calf Breadth: The maximum horizontal distance between the knee width and the minimum leg width, ie the maximum horizontal width of the contour between the knee width S6 and the Z value of the heel S8;

Suppose ncalf = {i^. j P. e Qsf,z(Ss) < z(^) < z(S6)} 59 = MaxiHdistiQ^)) (12) s io hip height (Side Hip Height): From the lateral horizontal hip width position to the vertical height difference of the heel, that is, calculate the Z value difference from the hip width S4 height to the heel point Sm; 510 = z (54) - z (520) (13) Sn thigh length (Side Thigh Length ): From the point of crossing to the vertical height difference of the knee width, that is, the difference between the kneeling point F5. 〇 and the knee width S62Z value; 511 ~ ^(-^5.0 ) ~ ^(^6 ) 21

Where Pa&Qsf ,z(P ) = ^l〇iJ + z(^ ° 2 1315042 (fourteen) S12 Body 距 (Side Body Rise): the vertical height difference from the side waist height to the underarm point, ie Horizontal waist width S! Height to the subordinate point F5. Z Z value spacing; *^12 = ~ ^(-^5.0) (15) S〗 3 foot length (FootLength): maximum foot length, that is, in the smallest leg Below the width of s8, sum the maximum horizontal width of the heel point S2.g;

Suppose Ω/οο< ={i^. |/>. €Ωί,ζπώι(Ω,)<ζ(?/)<2(58)} S9=Max(Hdist(Qf〇〇t)) ( 16) S14 side chest ^XChestDepth): The lateral chest at the height of the lower abdomen is the horizontal distance between the upper and lower body curves of the F2.0L and F2.0R at the average height of the left and right lower points. S14=Hdist(Pa,Pb) ,Pi>Q^Sb ^(Pb) = z(Pb) In the calculation of the amount of deformation, the arbitrary amount in the deformation region, and the known point. :^^ Program 2 and each of the influences and 5 form a certain special = away from p, ^ point amount v, _ multiply p, as affected in = system ^ (卩)' will be points Change to the amount of change in ρ point, the proportion of the influence of Ρ,

Where 7 Ν(η)· thus any point Ρ can be obtained via a conversion function /(~令 22 1315042 corresponding point] where the power value k has different settings in different parts of the human body.

In the above invention, 'in the crotch around the hip area, because the camera takes the image by perspective projection' and the image processing produces an error in the contour process, so that the searched contour image is at a low position at the underarm point. The arm length L of the armpit point and the arm root circumference are maintained at a specific ratio r ··1 in the YZ plane projection length L′. Then, along the armpit ^, the inside of the arm of the contour pixel is taken as ^1/11 height pixel amount, each The multi-strokes formed by the pixels and the under-the-spot point are averaged by vam, and then the vector formed by each point along the outer side of the arm and the underarm point from the shoulder point is interpolated with iu, and the inner product is the point of the 〇. The line can get the ugly under the age of the shoulder and the galloping _ Wei up and down and ^ shoulder point (l + r) L; where 'the Γ: 1 to take the golden ratio 〇 618: i is better. In the above invention, in the hip-to-abdominal region, since the hip-shaped geometry of each person is not necessarily the same, the length of the photo contour curve and the model model at the thigh root between the hip circumference and the underarm circumference is not equal. If the relative position of the profile curve on the model model is directly made, then even if the contour curve of the 3D side view is suppressed, the mother will cooperate with the longer brewing department to cause three _ distortions, and find the unreasonable change in the abdomen. The method is an extension curve of the lower abdomen of the key side and an extension curve of the root of the thigh upward, and the two extension lines respectively fit the two quadratic curves as the I-line of the leg root and the lower abdomen. The extension line extends from the waist circumference and the point of the maximum f-curve value on the half-length curve to the side-down point: the total length of the ϊΐ is the point p to the lower part of the curved curve of the front half of the neck; The extension line can be regarded as the position of the wheel to obtain the "; if; type death;, the point is moved to the opposite length of the test, to obtain a better view.忒α秘 In the above invention, 'in the shoulder circumference to the underarm circumference area, the 32-arm amount of the awning plate must include the amount of movement of the front and rear centerlines: = The problem is how to obtain the arm from the side porch image Root circumference 23 1315042 To use the above process to complete the amount of movement of the front and the back of the neck before and after the deformation of the underarm and the lower part of the neck as the known point, the information of the face wheel _ can be deformed The position of the shoulder point, so the sigh is a known point' so the amount of movement of any-unknown point on the arm root of the model model can be obtained by these three points. Maxian is in the middle of the neck circumference to the shoulder circumference area, because it is impossible to search for the triangle from the front and the part: the neck circumference, the shoulder line and the partial shoulder circumference. Therefore, to calculate Deformation inside the area, solution point. Therefore, on the side profile, the position of the lower front neck point and the upper front neck point can be found in the area on the maximum value side, and the average vector formed by the lower back neck point is calculated under the shame calculation. ^pper^front neck^po'mt [·_,一乃气一--*

Num(Pi') is the lower back neck and the lower front neck, and the point shape and vector are satisfied.

Vflfeck ' ^LowerNeck ~ 〇3 upper and lower neck points parallel W - intersect with the second half of the curve to find, the front wheel _ axis __ Fan Yishu definition of the definition, as for the right right class so it is possible to pass the field The amount of movement at any point. The person can estimate the difference between the result of the three-dimensional digital body after deformation and the original human body. 4 county sand _ coffee (__mean_squarc>it two 24 A3l5〇42 deformation point group. There are also % points, you can write ^ A, = { k}f〇ri = \,2,^Na and the two corresponding points one) and one (the distance between the hearts is

' α,-)=*α;' ^=+(z; - ZJ The two RMS values, flat ·, maximum and minimum values are defined as RMS values: b(K Ta;y J{K k) Νλ «;) Average value: Maximum value: Minimum value: The method is three-dimensional human body measurement feature length, exhibition shape three-dimensional human small outsourcing ^ l.pft+1y The measurement result of the characteristic length, 25 1315042 Minimum outsourcing two Dimensional feature length 1 = 0 ε is the margin of each small segment around the feature, ". +1 table features the points on each small segment. [Embodiment] For the detailed method of cutting the book (4) and the wire and effect of the wire, the following is preferred. The implementation of Nahe _ detailed system is as follows, and then we will understand it in depth. The general flow of the charm ifrff example is as follows, Wei obtains the front image from the camera, and the point projection is also possible. After the series of ΐίί-value finding, the width can be only one pixel in order to establish a reasonable three-dimensional human body model. In this embodiment, the feature points, feature heights and feature widths of the production system and the old-fashioned system are as follows: ten feature points, nine feature heights, and thirteen features are broadly defined. Feature points, five feature heights, nine feature widths, ΐΐΐΓ ΐΐΐΓ ' 'domain and then three-dimensional 'frequency conversion (3Dm () rPhing), reference method i solution (10) bit ^! People's dry and dry template to establish three-dimensional deformation of mathematics The 维 儿 生 生 生 生 生 生 生 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片 相片The following is a detailed description of the methods and techniques used in the steps performed in this embodiment. Image pre-processing: Boundary and side photos 'The image of the image processed by the image processing has a grayscale image. Cong k, dosing method (Just reference wire 24), c 26 1315042

6) 'K.mean, JS to color image Although these methods can clearly distinguish the only piece of literature 1) Shadow and the complex background of surrounding images, from the = boundary 'but but have ^ ^ ^ automatically Identify which part of the contour belongs to the person in the binarization to remove the miscellaneous points (as shown in the fourth figure), step four • such as the first = kg?). Although the above embodiment is more complicated than the use of the old field, but because of the manual processing, even the more complicated steps can scale to the clear boundary of the human body, boundary search and sorting: ,, the digital photo is removed from the image by the image, binarized, and edge-finished. ί Go to the pixel-wide round of the auditory image, so that the pixel is not the contour to find the second U into the light f-curve method.靡 靡 靡 ί ί ί 以及 以及 以及 以及 以及 以及 以及 以及 以及 ί ί ί ί ί ί ί ί ί ί The related process is shown in the sixth figure. The basic unit of the digital image is the image, the image is searched from top to bottom, from left to right, and the pixel that is searched for ^ stops, which is described as some before sorting. It is stipulated that the variable infection is first used to record the direction of the current pixel search, and the initial value of dkecti〇n is preset to 3 'also ^variable C_ectNumber, which is used to record the number after pixel sorting. Save the pixel number of the check to ConnectNumber , Checking the surrounding eight adjacent pixels in a counterclockwise order, the value of directi〇n is updated to the first direction number in which the pixel is searched for 1, according to the direction to enter the next pixel, and 27 1315042 front pixel, silk ― is an odd number, then (―) divided by 8 to take the rest of the heart and then counterclockwise m to the next pixel, silk 1 _ number, advance c. One _er is 2, and: pixel, corresponding to - ber is !, then the sorting is completed == The method completes the step of the contour pixel point and the step of S complex ^. The third method of bending value: to determine the person Zhao: r two: = the large segment represents the bending point value of the point 'vector sum The position of the transgression may be the feature point 'because of the logical operation of the method, and! After four subtractions, two additions, and one judgment comparison, the method of calculating the curvature of the curve in the program is much simpler. The value side (four) piece is a group of contour pixels sorted according to the adjacent relationship. Please refer to the seven figures. Suppose the coordinate position of a certain pixel point, the vector formed by the contiguous kth snack, pfi *k = [Xfd (ι)^Υβ (/')) by 6 and its former k ^ Is configured vector

PlPi-k =(^w(0»^rf(l)) where 28 1315042

Xfd(f) = Xi+k -, ^fd (0 ~ y^k - yi ^bd (0 ^ Xi-k " Xi

Yb<s (0=*^/-a - Vi (0+zw (0^/, (0+3⁄4 (0) makes HaW+lWI, rH heart W+rwW| then defines the variation of e point For BV(P) = xa^{Xc,Yc) (1) The decision of k value affects the result of the bending value of the point, so the choice of k value will have different settings in different sections, and the empirical value shows the comparison of common features. It is not obvious that a smaller k value is given, and a larger k value is used instead.

4. Deformation method: The deformation method adopted by the present invention is a deformation technique of a three-dimensional model proposed by Shepard in 1996 (see Attachment Reference 9), which assumes that the target of the deformation has a number of known points, and the movement of the unknown point is According to the distance from all known points, it is used as a variation parameter to determine the amount of movement of the unknown point. However, the present invention deduces a mathematical equation that facilitates the calculation of the amount of deformation, so that the operation processing speed is wide. As shown in the figure, it is assumed that the rectangle is the original shape, trapezoidal, and visible. The ti table is not at a point known on the target shape, Gi represents the point on the original shape corresponding to t 29 1315042, then the vector "_.· indicates the known point, as shown in the figure y, at any point in the area. The distance is ^=redundant. Suppose p point is affected by each 〇, and $' is known as the number relationship iv(6) between points ,, then multiply each point change by '乂5 to form a specific function influence. The proportion of the force, the change of the eve point ', ', and again, affects all pairs of p Σ ^ ·) (2) /»1 where ri U) therefore any point; 7, can be via a conversion function After m=p+±dt (3) in the different parts of the human body in the different parts of the human body, the pair is in the sense of the front and the side of the wheel it ί 4 Ρ 2' Ρ 5 9 ? ' 'hypothesis two: circumference = ', tenth In the figure, the formula (4) (4) is known as the point = 壬 is 2, and the example of p is between the materials, and the formula is as follows. '1 6 A, P«6, A, A ' then calculate its tf(*P + di+d6+dui+du6« (5) 30 1315042 丨』,ΟΣ>(0 Σ^·> dn^N ^!>w d.. d„ m Σ^·> i HUf νΐΓ^κ) Σ^·) In the above formula ^Nir,) =N(r,) + N(r6) + N(rul) + ΛΓ(γβ6) + N{rn) + N(rl6) V. Deformation evaluation: The embodiment of the present invention evaluates the results of the three-dimensional digital human body after deformation and the original dissimilarity 'in addition to calculating the cross-sectional area of a certain number of characteristic enclosures The length of geodetic measurement ΪίίίΧ, also refer to 1CP (interaetiVedQ-X see attached · Calculate the distance corresponding to the two points group, and take the scan = RMS (r〇〇t-mean_Square) value, the average evaluation criteria, the calculation method is as follows Take the hole 伹 /, dare value as the three-dimensional number of people before the deformation of the _ point group for the practice of a point, you can write can be written as a deformed point group τ also has a number of points, N, define two corresponding The distance from ~Μ) is 31 1315042 d(s,^)=13⁄4 ^-1=>iw-^)2+u-^)2+(z;-z.)2 (6) The 8 values of the abdomen, the flat touch, and the maximum health minimum order are defined as (7).

RMS value: D Average value: b(c -I) (8) (9) Maximum value: (10) Minimum value: The process of human measurement using the ίί use case. The length and degree: including f length can also automatically capture the human body's two-dimensional features. Some known two knee-high high data, according to which to build a model similar to Wright. Obtaining the corresponding shrinkage must first be defined for the definition of the human body two-dimensional measurement project document 30) (see attachment ir^xfr,: based on, build 4: meaning, reuse geometry definition, and —妓' dimension 2===== 32 1315042 Dimensional measurement results of human body features. The following is an explanation of the individual two-dimensional measurement items. The characteristics of the human body are shown in the twelfth figure. , human body side wheel _ levy size ship. Now and tenth. Then 'describe the definition of point set and image coordinate representation.

Ω/ front view outline point set. Ω, = Seto 〇;,, 2,.) 丨 Vi e W Ω, side view contour point set. ω: = ) I w € w Bu (Liu, 呢), 咐)): A contour point. , 冲), z(8): the γy ” component of the coordinates of point 6. y(F), z(i〇. The Π dimension component of the feature length of the front view. The outline, the willow. The side of the feature length S where the feature length S is located ” Dimension component W Μ contour point, 纟 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

_(Ω/) ··The total number of front contour points. ΛΓΜ/μ(Ω,): The total number of side wheel porches. Corpse cLroW · The shape of the front view. ^centroid · The centroid of the side view.

33 1315042 ^ίη(Ω) = {M«W·))1^-eQ} ^(Ω) = {^(^·))Ι^€Ω} ^η(Ω) = {Μ«(^.) )|Ρί6Ω} ^(^) = {Mm(z(^))|PfeQ} ~(Ω) = { Σ(· Nunt(Q) |“Ω}

Hdist(af) = {| yiP^-yiPj) \ \ € Q/sVz, 3(^.),ζ), αηά{γ{Ρ}\ζχ t ^ j}

Hdisms) = {| ^.)-^.) 11 Pj 6 Qi)Vz>3Wi,Xz)> and(x{Pj) z)i . ^ j} Vdist(Pl,P.)=\z(p .)-z(pj)\

Hdistnf (/>,/>.) =| yiP^-yiP.) |

Hdist^ (/>, p.) =( x(<p_) _ x(^p_) j

Dlstaf{Pi,Pj) = ^[\ Z(p.) - Z(Pj) i2 +1 istns(Pi,Pj) = ^j\ z(P.) -z(Pj) \2 +1 -T^ ^yjT Detailed description of each feature of the Congxing's feature can be referred to the text derogatory 'reference contour method body measurement project (see attachment, based on the basic definition of the solution i mathematical representation) based on the geometric definition In fact, one of the inventions is to derive a mathematical equation to measure a number of two-dimensional feature items. The generation of such mathematical equations is that the invention can quickly complete the two-dimensional feature two-item project, and further Quickly obtain the amount of deformation to correct the model model, so that the key to the pseudo-dimensional digit human body model, so the generation of these mathematical equations is also the characteristics of the month = ... respectively, the following are on the front and side contours疋One-dimensional feature item and the mathematical equation to be obtained·· First, the amount of projection available on the front side φ F1.0 Head apex (HeadPoint)!·The highest point on the human head. 2. The maximum Z value on the contour of the human body Point 3. The point where the 'Z value is the largest in the rim. If the same value of 2 is not only the average value of the point is the head apex. Take all

Fi.o meets the following conditions: 矸〇εΩ, Αο = {〇〇;, 2)丨V,, 2 = 2 jobs (Ω,)} ΛΜίί,ο) (11) if Num(Fl〇, M, then y (F'〇, = /Num(F'〇)

Head width (HeadBreadth) 1. Above the ears above the ear, the maximum horizontal line distance. The two sides are corrected, and the left and right sides of the 4th section are the most 3. From the height of the head apex, the first horizontal width is searched for a negative position. The width is recorded as the head width. Ten visibility and the next rate of change

/ - {ΗΛ5ί(Ρβ5Ρ6) | pa e e = 3/, 3 1-2_/; +乂_1<0 F^Max{z(f>^ 1315042

FzolFzor (left and right) points down (human claws ^? to nt) 1. The uppermost point of the left and right ankle sockets. 2. The intersection of the left and right inner arm curves and the torso. 3. ίΐ. The point group starts from the point of the smallest γ value to the point group between the minimum z values, and finds the large Z worth the left point F2.GL. The right half point group starts from the point of the largest γ value to the point group between the minimum Z values to find the maximum z worth the right side of the point.

Suppose , ={/>(/>a</<Aa = m^^(Q^))5>5 = m^(Zw. 々也(13)

Suppose Ω^(ν «<ί'<Α« = ^βχ(>;^(Ω^))9^ = Ι>Ι^^^Ω^^ => z ) =, (Ω0-ftJ ?) (1))

• F2 neck width (NeckBreadthJ 1·about the seventh cervical vertebra, the horizontal distance of the neck. 2. Below the head and below the armpit, measure the shortest horizontal straight distance between the left and right sides of the neck. 3. In the head width F! The height is between the height of the left shoulder point F3〇L, and the shortest distance between the contour points of the left and right body of the human body at the same level is obtained.

Suppose = {^ | € Q^ziF,) >z(i>)> z(F30L)} =>F2= Min(Hdist(Qneck)) ( ^) • FiolFior (left and right) shoulder points ( shoulderPoint) 1. The outermost protruding point at the junction of the left and right shoulder blades and the clavicle. 2. The point where the curvature of the left and right shoulders is the largest. 3. In the left half-point group, the point group whose index value is between the head vertex and the minimum γ value respectively finds the point and the same point with the Y value of the left sub-point F2.0L, and is here. Find the maximum BV value between the two points to get F3〇L. In the right half of the group, the point group between the point where the index value is between the head apex and the maximum γ value, respectively, 36 1315042 finds the point with the Y value and the same z value as the right squat point F2.〇r, And find the BV value between the two points to get the most. &<ΐ<β,

Suppose ^〇^^a<ind^F2〇Lyy^ z{fi)~z{F2 ) ^>F3XiL=Max(BV(P^)

Suppose ^lrighl_shoulder = Ρι\Ρ^ηβ, α<ΐ<β, a > index{F20R), z{a) = z(F20R),fi > mdex(i?^y{p) = y{F2_0R ) (17)

And find (19) ^FyoR=Max(BV{P>j) • F3 Shoulder Breadth 1. The lateral distance of the shoulders when the arms are placed naturally. 2. The linear distance between the ends of the left and right shoulders. 3. First calculate the left and right shoulder points F3.〇LF3〇R on the contour points of the human body, and find the straight line distance between them.

Fz=dist{F^L, F30R) (18) • F4 Bust Breadth 1. The chest width is the distance between the left and right sides of the chest. · 2. The straight line distance between the left and right points. 3. First calculate the straight line distance between the left and right points of the human contour point F2 on F2QR. F^dist{F2〇L, F2〇R) F5.〇 Crotch Point 1. The top of the crotch. 2. The intersection with the highest curve in the left and right legs. 3. Find the point with the largest value of 2 in the curve of the left and right inner legs of the human body. From this point to the 1/30 height range, the middle point of all the sorting points of the curve in the leg of the range is sought as the squat point. The curves in the left and right legs are defined as: 37 1315042 The curve formed by the group of points between the lowest point of the right half of the body and the lowest point of the left half of the body under the condition of point contour sorting. &<ΐ<β,

Suppose Ω 'inside a legiine a = Min {index{Znia (Ω^)), indexiz^ ))}, β =Max{index(z,^ (Ω^ ))}

Suppose Ω _ point

Pi \ Pie Qi"Side_ieg/i„e,a &lt; i &lt; β, <ind as Huma^ β &gt; index(Zttm (aM^e)), Hdist{fi,z_ {^^)) = HjmanHeight &lt ;a

Yiindex{〇lcr〇tch. Fs.° Num[Q—) (20) Reference F5 Front Waist Breadth 1. When the human body naturally breathes, measure the shortest horizontal straight line distance between the left and right sides of the trunk between the hip bone and the rib. 2. Between the lower point of the armpit and the height of the armpit, lay the shortest horizontal straight distance between the left and right sides. 3. The Z value of the lateral waist width Si corresponds to the waist width of the front rim.

PaeQJL, PbeQJR, z(Pa) = z{Pb) = Z{SX) => F5 = Min{Hdistaf (Pa, Pb)) (21) • F6 Natural Waist Breadth 1. Natural breathing At the time, between the hip bone and the rib, the small circumference of the trunk is measured. &amp; 2. The shortest straight distance between the left and right sides of the trunk in the middle of the height of the squat point and the squat point. v' 3. Find the shortest distance between the contour points on the left and right sides of the torso between the point of the Fu value of the FuFm and the height of the Z value of the lower arm.

Suppose nnear wibmne = {(/&gt; I /&gt; 6 Ω/, ζ(^)ρ_(5: hour 2 &gt;Z(i&gt;)&gt; z(F50))} =&gt;F6 =Min{dist {€lnear mmne)) (22) ♦ F7 HipBreadth 1. The maximum frontal projection around the trochanter of the buttocks when the body is standing upright. 2. Below the horizontal waist width, above the kneeling point, the maximum horizontal line distance between the left and right sides of the torso.

3. The Z value from the side hip width S4 corresponds to the hip width of the front contour.

Suppose y(Pa)^y(Pb), y{Pe) = y{Pcentroid), = = z(Pc) = z{S,) e fl. eQy F7 = Min{Hdistaf (Pa,pc)) + Min (Hdista/ (Pb,Pc)) (23) • FslF8r (left and right) Thigh Breadth 1. When the human body stands upright and there is no muscle tension, it is highest on both sides of the left and right thighs. The horizontal straight distance of the position. 2. Cross the point to the horizontal line distance between the left and right sides of the thigh (left and right). 3. The horizontal width of the left and right legs of the same height as the underarm F5.0.

Suppose ^n€ar ^high = {p, I y{Pd ^ y(F5_0), z(Pt) = z(FJ0), ^ e ) ^Fsl =Min{Hdistaf{F5Si,Pa% Pa eQnearJt;fttliigh ( twenty four)

Suppose ^near righlthigh - I y(F^) φ y(FS0), z(i^) = z(FS0\p. e | = Minimise, Pb% pb, anear righmgh (25) • F9L F9r (left and right) Low Knee Breadth 1. The horizontal width of the lowermost edge of the kneecap. 2. The horizontal spacing at the bottom of the knees of the two legs is β 39 1315042 3. Search for the inside of the left and right legs from the underarm F5,0 The maximum BV value between 1/8 and 3/8 height length and the f靡 curve opening in the +Y direction (f9L) and the γ direction (Fpr) are obtained, and the left and right knee horizontal widths are respectively obtained. .

Suppose = {Pt I Pi e Qinside jegiine! } Dirty (Ω/) + |Ζπώι(Ω/)&lt;咐)&lt;(|ζ·(Ω/) + ^‘(ω,))}

Let k = 35=&gt;^Pi^^inside ]eftIegline3Pa 3BV(Pa)&gt;BV(P.) F9L = Hdistaf (Pa,Pb), Pb e (26) where y(Pa) ^ y(Pb), z(Pa) = 2(Pft)j

Suppose ^ ι Ω^} 1 7 (-^(Ω/)+-ζ^(Ω/)&lt;ζ(ί&gt;)&lt;(|Ζη)κ(Ω/)+5^(Ω^)}

Let A: = 35 =&gt; Vi^. e B Pc 3 BV(PC) &gt; BV^P,) F9R :Hdistn,{Pc,Pd,,Ρά and (27) wherey(Pc)^y(Pdl z (Pc) = z(Pd)9Pc9Pd gQjr • Fiol Fior (left and right) Mid Thigh Breadth 1 · The body stands upright, no muscles are tight, 'lower to the knee midpoint, (left , right) the horizontal distance between the left and right sides of the thigh. 2. The horizontal distance from the midpoint of the knee width to the (left and right) knees. 3. The z value of the height midpoint between the lower point F5.〇 and the low knee width F1SLF15R respectively Take the horizontal width of the left and right thighs. ^xol = Hdist(Pa, Pb) ( 28 ) where y{Pa) ^ y{Pb), z(Pa) = Z(PJ = £(/s.〇) + 4^sl), p 2 戽(29)

Fior =Hdisi(Pc,Pd) 1315042 where y(Pc)^y{Pd), z{Pc) = z(Pd) = .ziF^.tz^m\ Ρ&gt;ρΛ&amp;ςιβ •卩11.〇1^ 11 dishes (left, right) heel point (; therefore 61? 0 shirt;) 1. The left and right tibia innermost edge of the most prominent point. 2. In the left and right calf to sole curve, the point of maximum curvature. 3. From the lowest point of the left and right half-length to the upper ι/8 height, in this range, find the maximum BV value in the left and right legs.

Suppose anear Hlheel = {/>gt; I € ^imideMaegUne,z{P^ &lt; (^ηΒΧ(Ω/)+|ζηιίη(Ω/))}

Let A: = 35 =&gt;V/&gt;eQ(W(fe w/ne 3 〇£ =Max{_BV{P^) (30)

Suppose nnear_rtghtieel = {/? I λ:8Λ//^,2(/&gt;) &lt; (^ζηΗΧ(Ω/)+^-ζΙιώι(Ω/))}

Ld k 35 ^ ^inside^rightlegline 3 FU0R=Max(BV(I&gt;)) (31) Lu? 111^11 厌 (left, right) calf width (〇 &amp; ugly tear (1111) 1. The body stands upright, the legs are slightly separated, and the weight is evenly distributed between the legs, the calf (the tibia) The maximum horizontal distance between the left and right sides of the edge. 2. The maximum horizontal distance between the left and right sides of the calf.

3. In the left and right low knee width F9LF9r to the heel height of the heel Fh olFh.or, find the large horizontal width of the left and right calves respectively.

Suppose ^ = {(Pi I P.t G Q^,z(F9i) &gt; Z(P.) &gt; z(F110J)} F1U =Max{Hdist(aneor 1φ〇ι/)) ( 32 )

SupposeQeear righlcalf = {(/&gt; I P.t e Ωβ, ζ(^) &gt;2(/&gt;)&gt; z(FUOJ?))}

Fn, =Mox{HdisHClnearrightcalf)) (3)) • F12L F12r (left and right) Min Calf Breadth 1' The body stands upright, the legs are slightly separated, and the weight is evenly distributed in the two 1315042 legs. The minimum horizontal distance between the left and right sides of the calf (below the upper edge of the humerus). 2. The minimum horizontal distance between the left and right sides of the calf. 3. In the left and right lower knees, see the F9LF9R to the heel height of the heel Fu.olF 11.OR, and find the minimum horizontal distance between the left and right calves.

Suppose \Pi^QJL,z(F9L)&gt;z(Pi)&gt;z(FU0L))}

Fl2L =Min{Hdist{€lnear leftt[iacalf)) (34)

Suppose ^near righlniacalf = {(Pt IP. € Ω^,ζ^) &gt;z(^&gt;&gt; z(FU0R))} = (Liang ί(Ω·呦,一,,)) (35) ♦ Fi3.〇LFi3.〇RCi,S&gt;te^a(NeckPoint) 1. The intersection of the left and right neck cap muscles and the shoulder line. 2. The most curved point of the left and right neck. 3. Between the height of the neck width F2 and the height of the shoulder point F3.〇LF3.〇R, find the point where the BV value of A=20, h35 and b50 is the largest, and take the average of the three points.

Suppose Ω_ =|/&gt;|i&gt;eQ^&gt;z(F3.0J&lt;z(^)&lt;(^2_(Q/)+^zmin(Q/))| ^ VPc〇η F _ Mu( 5F(i&gt;))L20 +Max{BV{Pi))\k^ +Max{BV{P^\ I near lefineck JJ 13.0X —* Γ —-- (36)

Suppose ^near_righmeck = I /&gt; e Q^ziF^) &lt;&lt; (^^(Ω7) + ))| VPc〇Ί F . MaxiBV^) |t=20 +Max{BV{P^) |i =35 +Max(BF(I^.)) L ,rv^t^^near nghtneck d Λ3.ΟΛ ~------^^0 (37) Spring F13 Lying long (TrunkLength) 1. The body is quite The total length of the body torso when standing straight. 2. The difference in height from the average side neck point height to the underarm point. 42 1315042 3. The left and right side neck points F^.or average height to the cross point &amp; 〇 z value gap F = Z (^13.0 £) + 2 (^〇^) 1 2 1 5〇) (38) ♦ FM Hip Height (HipHeight) 1. The vertical distance from the horizontal line of the hip width to the heel when the body is standing upright. 2. The difference in height from the hip width to the heel. 3. The hips see the F7 level to the heel point f11 () lf11 〇 r average height of the z value difference. ^4 = zC^li.〇£) + zd〇j?) 2 - Ming) (39)

• Fi5lFi5r (left and right) Knee Breadth 1. The knee width of the height of the left and right cheekbones. 2. At the height of the humerus, the horizontal distance between the points on both sides of the left and right legs. 3. Find the maximum length of the 1/8 to 3/8 height between the left and right legs by the Fm point Fm and the contour curve opening to the +Y direction (F15L) and the Y-direction (F15R) BV value, respectively, the horizontal width of the left and right knees.丨 e an I Ω,

Suppose Qiwjae Jeftlegline (g Zmax (^/) + — (Qf) &lt; z(i]) &lt; (- (0^.) + - Z^B (Qf))

F\5l = Hdistaf, Pa, P^), Pb 6 ζϊ凡 (40)

Wherey(Pa)^y(Pb),z(Pa) = z(Pb),PaJPb^ilJL I G 〇&gt;insi4e jegiine I ,

Suppose Clinside rigMiegiine 4^(Ω/) + ^π»η(Ω/)&lt;ζ(/?)&lt;(|ζιη3Χ(Ω/) + |ζπ,η(Ω/))

Let k = Z5^&gt;yP^Qinside^ghtlegline 3 P^BViP^BViP,) 43 (41) 1315042 where y(Pc) ^ y{Pd ), Z(PC) = z{Pd ), Pc, e • Fi6L Fi6R (left, right) Knee Height 1. The vertical distance from the knee width to the heel. 2. The difference in height from knee width to heel. 3. Left, right knee width Fi5lFi5R^J^ Pingnan degree to left and right heel point Fii.olFii.or Z value gap. (42) ^16« = Z(^ll.O/t)~Z(^lSJl) (43) Spring F17 Scye Deep (Scye Deptii) 1. The vertical distance from the side of the neck to the level of the underarm.

2. The vertical height difference from the side neck point to the underarm point. 3. Left and right neck point Fi3.GlFi3.GR connection 'F2.GLF2.OR line left and right lower abdomen, take the Z value difference between the midpoints of the two links. (44)

Fn= Vdist(Fii0L +Fuor , F2 0L +F20r·) • Fi8 Thigh Length 1. The vertical distance from the underarm point to the horizontal height of the knee. 2. The vertical height difference from the kneeling point to the knee width. 3. 胯 点 point F5. 〇 to the knee width F15LF15R average high Z value gap. Ρη=ζ{Ρ5〇)_^ψΕηά (45) • Fi9 Height (Cotch Height) 1. The straight line between the underarm and the heel when the body stands upright. 2. The vertical height difference from the kneeling point to the heel point. 3. Under the neck point F5.0 to the left and right heel points Fii.olFii.or the average Z value difference. F' — F5.〇) + Vdist(Fu.0R, F5.q) ( 46 ) 1315042 F*2〇 Shoulder Height 1. The vertical height from the shoulder to the heel when the body is standing upright. 2. The average of the two vertical height differences from the shoulder to the heel. 3. The average height of the two shoulder points Fwl F3.0R to the two heels Fu 〇lFii 〇R average high sound

The Z value gap. X 2 - (47) F21 NeckHeight 1. The vertical height difference from the side neck to the heel when the body is standing upright.

2. The difference in vertical height from the side neck point to the heel point. 3. The average height of the neck thief F from both sides to the Z value of the average height of the heel fu 〇lFu muscle. 2 (48) F22 BodyRise: i. The vertical distance from the waist height level to the underarm point. 2. From the horizontal height of the waist to the vertical height difference of the underarm point. The horizontal waist width F5 level to the z point of the underarm point f. ^ = ^s)-2(F,0) ·

The amount of projection available on the side of the Si Si. S Head Head Point The highest point of the human head. 2. The point where the Z value is the largest on the contour of the human body. 3. In the contour, the point with the largest Z value. If the same z value is not only one point, take the middle value of all points as the head vertex. ^10 eQ/&gt; ^〇={^(x,z)l Vi, ζ = ζαακ(Ω1)} Peach.〇)

&quot;师(uy (50) if Num(Sl0)^:i,χ(υ= η 45 1315042 • S S Side Waist Breadth 1. When the human body naturally breathes, between the hip bones and the ribs, The shortest horizontal distance from the waist to the abdomen was measured. 2. The shortest horizontal spacing between the front and the back of the body in the middle of the height of the inferior and inferior points. 3. The average of the F2.0LF2.0R in the lower part of the armpit Between the height and the height of the squatting point ρ·5〇, take the point group whose z-value is between 1/2 and 3/4, if the faceright mussel 1J 'takes the coordinate point of the smallest X value, if not Face rightO, take the horizontal distance between the coordinate point of the maximum χ value and the curve of the precursor.

Suppose ΩηωΓ = {P. IP. e Q

(-(Ωί ) + - (Ω, )) &lt; z{P.) &lt; (^ (Ω, ) + (Q, ))} near _ badcwaist if face_rightQ, x(Pa) = Max{x{P .)\ /&gt; e Ω if not face_right〇, x(Pa) = Pt 6

Sr = Hdistni (Pa, Pb), wherez (Pb) = z (Pa) &gt; Pb (51) Lu S2_〇 Side Heel Point 1. The most prominent point on the medial lower edge of the humerus. 2. In the calf to sole curve, the point of maximum curvature.

3. Take the minimum Ζ value of the contour to the bottom 1/8 range of the maximum Ζ value of the contour ‘ to find the maximum BV value of the back body curve.

Suppose 〇„ear_sidehee, = {/^. IP. e Qsb, (Ω,) &lt; ^(/^) &lt; (_ ) + Z^j

Let sideied 3 520 =Mxx(5F(/&gt;)) (52) • Human Height 1. When the human body stands upright, the straight line from the sole to the top of the head is shown from the side view. 2. The height difference between the heel point and the head apex. 46 1315042 3. From the heel S2.0 to the top Su is at the Z value spacing. =z(Su0)-z(S2_0) (53) • S3 WaistHeight 1. The vertical height difference from the side waist width to the heel when the body is standing upright. 2. From the side waist width to the vertical height difference of the heel. 3. The horizontal waist width is 31 to the heel point S2. =z(5,)-z(520) (54) ♦ S4 Side Hip Breadth 1. When the body stands upright, it projects around the largest side of the hip rotor. 2. The boundary is below the horizontal waist width, above the kneeling point, the longest horizontal straight line distance between the curve of the front body and the curve of the back of the body. 3. The point group between the side waist width Si and the Z value of the underarm point F5.〇, if face right〇, take the coordinate point of the maximum X value, if not faceright(), take the coordinate point of the smallest X value , the horizontal distance from the curve of the predecessor. ^near_sMehip ={^1^6 Qsb, z(5,) &gt; ζ(ί)) &gt; z(Fs〇)} if face_right〇, Pt ^ if not face_right〇, x(Pa) = Min(x( P;)), P{ ^^near_sidehip S4=Hdisicl (Pa,Pb), where z(Pb) = z(Pa), Pb^〇sf (55) Spring S5 Leg Thigh Breadth 1. Straight standing when there is no muscle tension, the horizontal width at the highest position of the thigh. 2. At the height of the squat point, the horizontal width of the sides of the thigh. 3. Calculate the horizontal width of the two contour points relative to the height of the F5.0 and the Z value at the front. 47 (56) (56) ^S^Hdistns(Pa,Pb) • S6 knee width (8) also Knee Breadth)

(57) (58) 1315042 ^P〇,e {Pa,PJ^)^(P6), = 〇), P〇eQs/,p4e^}, 灭灭_立' in the wealth of the degree #转2, The horizontal width of the side of the most prominent point on the knee. · ^ Ten see distance.

Oppose |ρ0,Ρ6μ(Ρ0)^Λ(Ρ,), 2(PJ = z(PJ = £i^kl±f(5H) 0 . 1 2 ,弋s£VP6eft!6 ^S6=Hdistn(Pa, Pb) S7 Mid Mid Thigh Breadth L The body stands straight and stands sideways, when there is no muscle tension, the height of the squat point, the big _ _ horizontal wide distance 2. The height from the squat point to the side knee width The horizontal width of the point. 3. Calculate the water *re 4/X/U z(/&gt;a) = z(/&gt;4) = relative to the two contour points of the leg width F(10) and the Z value height in the front face. z(f muscle), /> 〇e, corpse e

Ss Min Calf Breadtfi 1. The minimum horizontal distance between the contours of the lower leg (below the upper edge of the cavity) when the body is standing upright and the feet are close together and the weight is evenly distributed between the legs. 2. The minimum horizontal width between the knee width and the lowest point of the contour. 3. Find the minimum horizontal distance between the leg width S5 and the Z value of the heel S2.0.

Suppose acalf = {Pi j Pi e nsf,z(S2〇) &lt; Ζ{Ρ{) &lt; z(S5)} sg = Min{Hdist{acaff)) 48 (59) l3l5〇42 S9 calf width (Side Calf Breadth) 1. When the body is standing straight and standing sideways, the feet are close together, when the weight is even between the legs, the calf (below the upper edge of the humerus) wheel _ maximum horizontal distance ^ two 2. Between the knee width and the minimum leg The maximum horizontal distance of the wide outline. 3. The maximum horizontal width of the contour between the knee width S6 and the Z value of the heel Ss. ^pp〇Se = {P. IP. € Qsf,z(Ss) &lt; Z(P.) &lt; z(56)} (60) Spring s 10 Hip Hip (L) High straight side of the human body When standing, the horizontal line of the side hip width is perpendicular to the heel of the heel. 2. From the horizontal hip width position to the vertical height difference of the heel. 3. Calculate the z-value difference between the hip width S4 height and the heel point s20. Hey. = paper) - paper. (61) • S 〗 1 side Thigh Length ^ The vertical distance between the height of the knee and the width of the knee when the human body is standing sideways. 2. The vertical height difference from the kneeling point to the knee width. 3. The difference in the Z value between the point F5.G and the knee width S6. • Hey, 5. )—Paper) (8) I2 Cross Body Rise 1 The horizontal distance from the lateral height of the lateral side of the human body to the height of the underarm. From the height of the side waist to the vertical height difference of the underarm. 3, waist width &amp; height to the point of the kneeling point f5 〇 z. • Γ milk. (63) FootLength L The horizontal distance from the tip of the talk to the heel. 2. The maximum length of the foot is long. 3. Under the minimum leg width &amp; height, sum the heel point S2.0 to the maximum horizontal width of 49 (64) 1315042.

Suppose Ω/οοί = {Pt I e Ω,, Ζπώι (Qs) &lt; z(Pi) K 2(5g)} S9 =Max{Hdist(Q.f00t)) • Chest Depth 1. From The thickness of the chest measured from the side. 2. The side of the knee is at the height of the chest. 3. At the average height of F2.0L and F2.0R at the left and right points, the horizontal distance between the curves of the lower body.

S"=Hdist(Pa,Pb) is said to be 2~bwb)=z(Pb) (65) In this embodiment, the contour of the human body phase is obtained, and the two-dimensional feature is obtained according to the two-dimensional feature information, and the weight of the three-dimensional human body model is performed. Therefore, the method and necessity of the i P root template model in the XY plane bidirectional extremum and the necessity j = region respectively explain the step of deformation.

Find the front profile before the weight; some of the information is insufficient = according to ======== is the geometric information of the pair. The outline of the miscellaneous part is S === the extreme value of the phase of the coffee, so that these poles are moved to the contour phase = 50 1315042 as shown in the fourteenth figure 'Defining the human cross section as the χ γ plane, due to the station Pose

® t &amp; search and subtraction on each of the "® direction" and 4 ^ I shot to get four points, set these points as the three-dimensional body deformation of the human body is approximately symmetrical, so in the upper and lower areas of the Lai and 胯The shape of the point of the structural point of the needle-fence heterogeneous model is obtained. The ρ is relative to the centroid in the γ square: water, the equidistant point ρ is as shown in the fifteenth figure, and the same method is used to calculate the relative value = centroid in Y The equidistant point of the direction? 3•' These two points are also added as known points. This area of the human body is flatter from the chest to the shoulder. The pattern on the model will make the X-direction extremely different from the lion. It is determined that this area is in the front center line of the human body. position. The x-direction of lying on the model model, the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ These two-dimensional known points will move to the opposite position on the line when the three-dimensional body is deformed. Next, the processing for moving the amount J at other points having the same non-extreme value is obtained by substituting the amount of movement of the known point into the formula (2). However, in different areas of the human body, the known amount of movement {is not necessarily obtained by the amount of movement of the extreme value, so in the following description, the order of the regions will be used to explain how to find the known points of the region. The amount of movement.

First, the armpit circumference to the hip area: The geometrical change of the human body between the underarm and the hip circumference is relatively difficult compared with other areas, and the unevenness of the surface or front contour is obscured, so it is relatively simple to search for known points. As long as the extreme value of each circumference is searched for four points of the known point, and then it is judged that if the extreme value of x does not fall near the t-heart line, then two points which are equidistant from the centroid in the γ direction are added as known points. Since the camera takes a perspective projection and adds image processing to generate a contour error, the searched contour image is at a low position at the squat point. As shown in Figure 17, this will result in a 3D deformation result. The settlement is so that the chest circumference is almost close to the under-milk garden to produce a distortion phenomenon, as shown in Figure 18. 51 1315042 Like, the next point is too low, this (four) reference in the cross-section of the arm of the projection of the arm of the cross-section of the length of the arm section only ten ~ length Z maintain a special ratio I &quot; : 1 , such as the combination of sound i = H is good, this adjusts the position of the point on the photo, so that the solution is closer to the actual one. According to the above definition, the shoulder point and the underarm point on the image of the photo rim are obtained, as in the twentieth figure ( a) As shown in the small figure, the lower point is taken in the wheel of the multiplexer, and the number of pixels is 1/n, and the number of vectors per pixel plate is averaged to obtain L. Then start from the shoulder point along the outer side of the arm: 腋?: The vector formed by the point and the inner product of the ^, take the inner product of 0 in the ten-figure (9) small picture), and the shoulder point (1 bamboo) L, so The results of the improvement are shown in Figure 21. Second, the hip circumference to the underarm area: the performance of each person ^ geometry is not the same, so someone's photo contour curve (as shown in Figure 12 (a) small picture) and the model model (such as the second In the twelfth (b) small figure, the length of the earning part of the hip _ 胯 不 不 不 不 不 不 不 不 不 , , , , , , , , 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度 长度The 3D side view has the same contour curve as the original photo gallery, but causes three-dimensional information distortion for the long thigh root, which causes the lower abdomen to produce unreasonable deformation results as shown in the small image in Figure 22. The solution is to first build · Bay ^ face belly, the extension curve of the round temple ^ and the extension curve of the thigh root upwards as the third === dashed line, so the virtual two extension lines can be applied in the lower abdomen and the thigh root respectively. The practice is to fit two quadratic curves from the known contour lines as an extension line of the lower abdomen of the leg root, for example. First, find the point using the maximum bending value method in the curve of the front half between the waist and the lower circumference (as shown in Figure 12). Similarly, between the ^ and the waist circumference, the maximum value of 52 1315042 is used to search for the household 2, and the household 3 is the middle point of 6 and /> 2, and the curve can be fitted by 66g, and the extension line is The length extends from the downward to the height of the lower neck. The curve r is from the middle, the 豸, the ,, the three points fitting the quadratic curve upward, which is the point of the south point of the squat point on the curve of the front half, which is the midpoint of the sum of $ and And the total length of the curve after the root of the leg is extended upward is m times, and the value of ^^ is preferably 1.3, which is a better result obtained by testing various lengths of the invention. The two extension lines can be regarded as part of the contour curve, and the extreme points on the model model are respectively moved to the opposite positions. Although it cannot be completed, the result is more reasonable, and it is shown in the twenty-fourth figure. Shown. ,

Third, the shoulder end is surrounded by the underarm area: In the shoulder circumference and the underarm area, if you want to know the deformation amount of any point ^, the known movement amount of the towel 纟 must contain the movement amount of the front and rear center lines and the front ^Half-arm root _ movement amount (as shown in Figure 25), the seed line is known by the side rim, so the main problem is how to get the arm root _ information, the hair 0 Following solution

I ^ is the deformation heart: Figure 27彳=bit; circumference::also = ΐ;;来=板模^ Fourth, neck circumference to shoulder circumference area: 53 1315042 To get the contour of the neck area of the image, you can On the model model, the X and Y * directions of the neck circumference are moved to the opposite positions on the wheel line. Therefore, on the side profile, the method of the maximum f-curve value (with a κ value of 3.5 is preferred) is obtained. The above area finds the next _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The resulting average vector v^edt is shown in Figure 29. · uPP^r_ front _neck _pomt imeck—Qiao. ·,— ^ Num{Pi) (66)

Then the vector of the next and the job meets the conditions

I I

Vfl4eck ' VLowerNeck = 〇 ( 67 ) g The back neck point is the same as the angle of the neck circumference from the upper front neck point parallel to the curve of the rear half body corridor. The lower right neck point and the lower left side Faced by the above definition of the first silk, as for the upper right ', the outline has no obvious features, so use the side wheel ^

ΐ ϋϋ ϋϋ 从 从 从 从 从 τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ τ It can be refined by the formula ϊ ^ ^ 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面 正面After the squat, shoulder line shoulder &amp; search 耑 求 求 求 求 求 任 = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = Table 54 1315042 This gives the sample model any point in the area; the amount of movement i of 7. In the course of the research, the present invention has been verified by taking four female models as examples. The numbers are 001, 002, 003, and 004, respectively, of which 00 002, 004

It is young and old, and 003 is middle-aged. In the implementation verification process, the front and side photos of four women were first obtained, and the human body was digitized after riding the body scanner. The number was also compiled as 〇〇 〇〇 2, 〇〇 3, 〇〇 4, and the model body. The model is possible, one of these four digital bodies. According to 〇〇1 as the model model, the three-dimensional digital human body model obtained from the deformation of the other three people's wheel sculpt is represented by 〇〇2@〇〇1, 003@001, 〇〇4@〇01, respectively. According to the foregoing research steps, the known feature lengths on the contour curves of 〇〇3 and 004 are respectively used to deform the 〇〇1 model, and the results of the shapes are sequentially displayed in the thirty-third figure of the thirty-third. In addition to the comparison of the model model, the wheel grinding curve and the deformation results, the figures show the results of the deformation in four different viewing angles, of which the thirty-third figure and the third-fourth figure are 0〇2. The deformation result (referred to as the verification example 1) 'Thirty-fifth map, and ~ two maps are the deformation results of 〇〇3 (referred to as verification example 2), the thirty-seventh and thirty-eighth Is the result of the deformation of 004 (referred to as the verification example three)

Because the present invention can reconstruct a three-dimensional body position model by using the definition of a two-dimensional ship such as various feature points, feature heights, and ultra-wide widths, and deriving reds, which can be reconstructed by the thirty-nine-theater. The human body digital model = and for the two-dimensional digital body to turn up a number of corrections, outsourcing, or the length of the segment can be further measured, for example, in the national people's body measurement number, or for example, the application of clothes, etc. It will be more convenient. Except, ^ Many industries or categories that can be applied are countless. The examples are as follows: 1. The development of personalized body shaping underwear should be 帛--not just lying on some physical integers, including arms and thighs. Weizui, this type of corset is a very personal product, so it is very convenient for users to make a uniform with only two photos. Α 2. Design and production of burnt and scalded pressure garments - can provide calorie-contained patients with a body size of 55 1315042 Custom-made pressure garments, pressure gloves, sturdy masks (scale items are more personalized), etc., and facilitate the development of pattern development Made with output. 3. Conducting national human body measurement 1 can alleviate the concern that the privacy and security of the body are violated when people enter the scanner space. As long as they agree to wear, the clothes (or tights) can take two phases&gt;{' It is extremely convenient to obtain highly reliable measurement data. Especially for the measurement of national human body in remote areas, or for those who are not allowed to go out to perform at home, it is more convenient. , the expansion of the human body bank database and the update of the work. The size of the domestic textile industry or manufacturer is used for design and development, such as uniforms, student desks and chairs, office furniture, furniture, household items, father tools, defense weapons, sports and fitness design, clothing Design and so on can be applied to the material core (10) anti-compliance equipment, build the most: further derivation (four) study equation to measure the human body

(0-eight (called +1)/2)|♦(6)-Zengwai!)/2) ί) Minimum outsourcing 3D feature long point '(===) section margin' W table features each The feature area on the small section, except: ί: the number of the difference in the amount of j-strips measured by two: two measurements on the amount of the knot, the ruler 0^2 'the width of the ground is the same), and (7) For the comparison of the styles of the towns and villages, the comparison of the magazines is based on the comparison of the maps of the state of the country. 56 1315042 (4) The small map is a comparison map of the small shoulder line, and the four mosquitoes (four) are recommended. The invention has the following effects and effects: 1. Defining the contour image on the human body: the old r ^ is defined for the corresponding position on the contour image. The description t defines at least + the description on the front contour. The method has at least two feature points on the side porch, and searches for and distinguishes the features of the automatic element into a parent, without relying on manual selection of two-dimensional measurement results. Since the present invention utilizes two photos, Can be r out: ΓΪΪ human body model 'more through the derived mathematical silk type to further ^ ^ f most shaken packet feature length, so that three-dimensional human body measurement knot It can be easily obtained for human body measurement, which is beneficial to many industrial applications, so that the accuracy of the industry can be improved, and the convenience of personalized products can be completed. The biggest contribution of Ming is that the two-dimensional features of the paste are reconstructed into three-dimensional complexes, and The approximation is quite immersive, and further can complete the three-dimensional human body's acquisition of the 3D human body broom, and the disadvantage of the tape measure and large error, the invention can process the three-dimensional 57 quickly and conveniently.

1315042 ☆ Body reconstruction and measurement 'For the people's measurement plan, human factor two j: Because the invention is deep == The patent is established; i see you = know, so it is completely invented [simplified description of the figure] The general flow chart of the embodiment of the present invention, the second figure is shown in the embodiment of the present invention. The background step is shown in the second embodiment of the present invention. The background removal step is shown in the embodiment of the present invention. In the embodiment, the eight-adjacent sorting method is performed on the pixel points of the corridor. The seventh figure is the schematic diagram of the bending value in the embodiment of the present invention. The eighth figure in the embodiment of the invention shows the original shape and the target shape before and after the movement. FIG. 9 is a schematic diagram showing the relationship between the present embodiment and the known distance. The tenth figure is a schematic diagram of the known points in the embodiment of the present invention. FIG. 11 is a three-dimensional human body change in the embodiment of the present invention. The human body in the embodiment of the present invention is positive ====. The thirteenth figure is shown in the figure of the present invention. The fourth embodiment of the present invention is shown in the fourteenth embodiment of the present invention. The fifteenth figure of the schematic diagram shows the hair In the embodiment, the sound of the known point is added. FIG. 16 is a schematic diagram showing the distribution of the two-pole value on the model model in the embodiment of the present invention. 58 1315042. The seventeenth item is the image of the rim image in the embodiment of the present invention. FIG. 18F is a schematic view showing the chest circumference of the contour image of the actual closure of the present invention being depressed. FIG. 19 is a schematic diagram showing the first embodiment of the present invention. Figure 10 is a schematic diagram of the method of retrieving the lower point of the wheel gallery in the manner of adopting the golden ratio in the embodiment of the present invention.

FIG. 12 is a schematic view showing the improvement of the chest circumference in the embodiment of the present invention. FIG. 11 is a schematic view showing the problem of the thigh root across the lower circumference of the embodiment of the present invention. FIG. 23 is a thigh in the embodiment of the present invention. Root Fitting Outline Extension Line Description FIG. 24 is a schematic view showing the improvement of the thigh root portion of the embodiment of the present invention in the embodiment of the present invention. FIG. FIG. 26 is a schematic view showing the outline of the arm root in the image in the embodiment of the present invention. FIG. 27 is a schematic view showing the known point on the arm root circumference in the embodiment of the present invention. The figure shows a schematic diagram of searching for the upper front point and the lower front neck point with the maximum bending value in the embodiment of the present invention. The twenty-ninth figure showing the lower back neck point and the upper back neck point in the embodiment of the present invention. FIG. 20 is a schematic view showing the upper right neck point and the upper left side point in the embodiment of the present invention. FIG. 21 is a view showing the shoulder end to the neck circumference area in the embodiment of the present invention, 59 1315042 The thirty-second map Illustrated in the embodiment of the present invention, a method for deforming the shoulder end to the neck region is shown in the thirty-third embodiment. The deformation result of the embodiment 1 of the present invention is not intended to be the invention shown in the thirty-fourth embodiment. The embodiment shows the result of the modification of the first example. The thirty-fifth figure shows the result of the modification of the second embodiment of the present invention. The deformation result of the verification example 2 is different from the perspective view. The thirty-seventh embodiment shows the modification result of the third embodiment of the present invention. The result of the deformation of the third embodiment is a schematic diagram of the three-dimensional digital human body model according to the embodiment of the present invention. The fortieth figure shows the structure of the three-dimensional digital human body model of the embodiment of the present invention. The line segment is not intended to be a schematic diagram showing the difference between the amount of Φ and the amount of tape measured on the three-dimensional digital human body model of the embodiment of the present invention. Annex: References a detailed aggregated

Claims (1)

1315042, apply for a patent garden: ^ The law '&amp; must be read by a computer as follows: '::, from the body of the positive and side photos of the motion processing, after:: the above-mentioned image of the hand-dimensional features =, surface Feature width two The model of the model is to adjust the two-dimensional digital human body model of the human body in the flat field; among them, the special feature; The definition of the calculations (4) and the equations for the calculations are as follows: the projection amount obtained by the front: (1) Fu) The head vertex (HeadPoint) on the human body corridor, the point of the largest z value, the same Z value is not only a little When taking the average of all points as the head apex; · FlA 6Ω/5^ι.〇= {Pi(y,z)\\fi, Z = (Ω7)} Step hist. 〇) / /Num{Fxa) if Num(FXXi)^\ , then y(F10) (B) Head width (HeadBreadth): above the ear above the ear, the maximum horizontal line distance, that is, the apex of the head Down the south, search for the first horizontal width and the next layer of change rate is negative, the width is recorded as the head width; / = {Hdist(Pa, Pb)\Ρα €eΩ^, 2(Ρα) = ζ( Ρέ)} Ύ&gt;3/Μ~2/ ί+/ ί-1 &lt;〇F^Max(z(f)) (2) 1?2.01^2.011 (left and right) 腋下点(入111^红For example: Left and right inner arms 61 1315042 Suppose ^r!ght_shoulder = PilPj e Cijg, a &lt;i&lt;β, a &gt; index(F2SiR), z(a) = z(F20R),fi &gt; Index(F2〇g),y(j3) = y(F20S) =&gt;F3.〇M == Max^BV (p.)) (6) F3 Shoulder Breadth: End of the left and right shoulders Straight line distance, that is, first calculate the left and right shoulder points F3.0LF3.GR on the contour points of the human body, and find the straight line distance between them; F,=dist(Fy〇L, FXOR) (7) F4 chest width (BustBreadth ): The straight line distance between the left and right points of the lower jaw, that is, the left and right points of the human contour point F2.〇LF2.OR are calculated first. And find the straight line distance between; F4 = dist (Fi0L, F2_0K) (8) F ^ o 胯 lower point (CrotchPoint): the intersection of the highest curve in the left and right legs 'that is to obtain the left and right inner legs of the human body The point where the Z value is the largest in the curve, from the point to the 1/30 height range in the -Z direction, seek the middle point of all the sorting points of the curve in the leg of the range as the underarm point; define the curve in the left and right legs as: a curve consisting of a group of points between the lowest point of the right half of the body and the lowest point of the left half of the body under contour sorting conditions; Suppose Ω, inside"egHne Ω/? α&lt;ί&lt;β9 a = Min [index{z^n (Ω^)), index^ (Ω β))}, hu = shib Φ — (Ω&gt;) ), iWexQ—(Ω 戽))} Pi \ Pi e Q〇iside_Iegnne,a &lt; i &lt; β, Suppose point =&lt;a&lt;mdex[ZmiX (QinsideJegline)), Hdist{a,zmilx (^insideJegli„ e)) = Humar^eighl β &gt; indexiz^ {〇.insideJegline)), Hdist[β,^)) = Humat^eiSh&gt; =&gt;f5_0 63 1315042 (9) F5 front waist width (H〇rjz〇ntai waist Breadth): between the lower point of the armpit and the height of the underarm, the shortest horizontal straight line distance between the left and right sides of the torso, ie the lateral waist width The Z value of Si corresponds to the waist width of the front contour; p〇&amp; ^jL, Pb eQ/R, z(Pe) = z(PA) = z(5I) Hin, Hdistaf{Pa, Pby) (10) &amp; natural waist width (&gt;^ plus melon 1\\^513 ratio &amp; (1 also): between the underarm point and the underarm point, lying on the left and right sides of the shortest straight line distance, that is, in the 腋From the point F2.olF2.or to the height of the Z value across the lower point F5.〇, find the shortest distance between the contour points on the left and right sides of the human torso; Suppose nnear waislline = {(i&gt; I p. e &gt; 2{ ^)&gt; z{F50))} =^6=Μ·(Ω_一)) (11) F7 Hip Breadth: Below the horizontal waist width, above the kneeling point, the left and right sides of the trunk are the largest The horizontal line distance, that is, the Z value of the side hip width S4 corresponds to the hip width of the front contour; Suppose y{.Pa)^y{Pb), y(Pc) = y{Pcentroid), z(Ptt) = z (Pb) = z(Pc)^Z{S4) pa eQf F7 = Min(Hdistn/ (Pa ,PC)) + Min{Hdistaf (Pb,Pc)) ( b) F8l F8R (left and right) Thigh Breadth: The horizontal line distance from the lower side of the neck to the left and right sides of the thigh (left and right), that is, the level of the left and right legs of the same height as the lower point F5. Wide distance; Suppose Ω^_1φ„ι?ι, = I y(^·)* y{Fsa\ z(Pi) = s Ω^} =&gt;FSi ^MiniHdist^iF^PJ), Pa ^€lnear ieftthigh Suppose Ω „Mail rtg melt igh = y(〇矣y(Fs.), ζ(β) = z(F5·.), /^ F, ^\QL 1315042 )FlR=Min{Hdis,,F5Q,Pbyh it^ near _rightthigh (13) Fsl FSR (left and right) Low Knee Breadth: the lowest level of the knees of the legs Spacing, that is, the maximum bv between 1/8 and 3/8 of the south length and the contour curve opening toward the direction (Fpl) and the -Y direction (FgR) are found from the lower left and right legs. Value, find the horizontal width of the left and right knees respectively; Suppose ={Pt fegffi(e I Ω^, (^^(Ω/)+|ζ^(Ω/)&lt;Ζ(/&gt;)&lt;( ^ζηΒΧ(Ω/)+|^πώ](Ω/))} Let k = 35=&gt; V/) e jefUegftne 3Pa 3 BV(Pa)&gt;BV(P.) F.^Hdist^P^P ^Q^ where y(Pe) Φ y{Pb), z(Pa) = z(Pb), Pa, Pbe Suppose= {P.\P. ^ I ^ (Ω/)(Ω/) &lt; z( i&gt;) &lt; (I Zj〇a (Qf)+l2min (Ω/))} :&gt; ^ _right/&lt;g/we 3Pe9fiF(Pc)&gt;5F(^.) F9R = Hdistar, Pc, Pd \ pd where y(Pe) Φ y(Pd), 2(Pc) = p^ g (fourteen) F10LF10R (left, right) Mid Thigh Breadth: kneeling down (left, right) knee width The horizontal line distance between the midpoints, that is, the underarm point and the low knee width fz value are divided into the left and right sides of the county; 15L where y{Pa) Ψ y{Pb\ 2(p ) = z( P ) = z(尸5.〇) + grab,) ° 2 Fl0S =Hdist(Pe,Pd) Pa&gt; Pb^ 65 1315042 where y{Pc)*y{Pd), ζ( ^) = ζ(Ρ,) = f(F^tz^)&gt; Qiao, Prf e Ω 2, 】 (fifteen)? 11_〇1^11 厌 (left, right) heel point (]^61?0 plus): in the left and right calf to sole curve 'the curvature of the point, that is, from the left and right half of the lowest point up about 1/ 8 height length 'in this range to find the maximum value of the curve in the left and right legs; Suppose ςΐ brim 丨 丨 = know \ Pi e inside ^leftlegline5 ^ (/?) &lt; (^ηΐΜ (Ω /) + 22π .η(Ω^))} Let k = 35 => eh ine 3 Suppose ^ear ^ed = {i^ | € ^^ute rightleglme^ Z(^) &lt; ^Ζ·TMχ (^/ ) + ~ ^ (Q/))} 2&gt;" = 35 =ν&Ω_ 3 Α10Λ=Μlike (5 which (16) FiilF11r (left and right) calfBreadth: the largest of the left and right sides of the calf The horizontal distance, that is, between the left and right low knee width F9LF9R to the heel height of the heel Fi1_OlFii_〇r, respectively, the maximum horizontal width of the left and right calves is obtained; Suppose Qnear J^ = {(Ρ{ I Ps &gt; z(^) &gt; z(Fn〇i))} Ful =Max(Hdist(Qnear ^)) Supposennear righlcalf = {(/ί I Pt 6 Ω^,ζ(^9Λ) &gt; z(^.) &gt ; z(Fn0R))} FnR ^ Mox{Hdist(Q.near rishlai¥)) (17) F12lF12R (left and right) Min Calf Breadth: left leg part left The minimum horizontal distance between the two sides, that is, between the left and right knee widths to the heel height of the heel Fu.olFu.or, respectively, the minimum horizontal distance of the left and right calves is obtained; Supposeζ1ηεαΓ ΐφΠάααιιχ = {(^- \Pt &^^,ζ{Ρ91^&gt;z{P.y&gt;z{Fll0iy)} FnL= Min{Hdist(flnaΙφπάααιν)) 66 1315042 Suppose ^nearfrightniacalf = {(^/ I e &gt; Z(^i) &gt; Z(^11.0i))} Heart = M« (Ma ί (Ω_—J) (18) 1?13.01^13 (left, right) side neck point (^〇1&lt;;?〇丨1 ^): The most curved point of the left and right necks, that is, between the height of the neck width F2 and the height of the shoulder point F3.0LF3.()R, respectively, looking for BV of k=20, k=35, k=50 The point with the largest value is taken as the average of the three points; Suppose ΩΛββΓ /^ =|/?|^€Ω/ι,2(^()Ι)&lt;ζ(^)&lt;(^ζηΒΧ(Ω/)+ ^Ζηώι(Ω/))| ^VPeQ 3 F - Μαχ(βν(ρ^ L^o +Max(BV(I))) |t=3S +Max(BV(Pi)) U Λ t ^ AAnearJeftneck ^ 1 I3.〇£ — -- &quot; 3 Suppose ^ = |i^ I /&gt; e Ω^,ζ^) &lt; z(P.) &lt; (±ΖπΒχ(Ω/)+ΑΖπώι(Ω/))| =&gt; vd a 3 gamma (4) U 3 (19) Fu torso length (TmnkLength): from the average side neck point height to the vertical height difference of the underarm point, that is, the left and right side neck points F ^ ol F13. 〇 r average height to the underarm point f5 〇 Z value gap; (20) Fm hip height (HipHeight): the difference from the hip width to the vertical height of the heel, that is, the hip width F? level to the heel point Fu 〇 L fu 〇 R average height z value gap; &amp;=υ+ Grab the path) (1) (2 _^1) fi5lfi5r (left, right) Knee Breadth: at the height of the humerus, the horizontal distance between the points on both sides of the left and right legs, that is, by the lower point f5 〇 along the left and right Find the maximum BV value of the inside of the leg between 1/8 and 3/8 height and the contour curve opening in the +Y direction (F1SL) and _γ direction (Fi5r), and find the left and right knee levels respectively. Wide distance; 67 1315042 S (^ΖπΒχ(Ω/) + ·^Ζπ»η(Ω/)&lt;Ζ(^)&lt;(^ΖTMιχ(Ω/) + -Ζηώι(Ω/))| Let k 35 ^ 运Qinsi ^一/j沙fjggiiffe 3 Pe bBV(Pc)&gt;BF(I]) ^\5r = Hdist^ {PciPd\ Pd g where y(Pc) Φ y(Pd)yz(Pc) = z(/&gt; ), Pc, Pd e (22) F16LF16R (left, right) Knee Height: Knee Height from knee width to heel' ie left and right knee width Fi5LFl5R level to left and right heel point Z Value gap; ^\6L = 2(^Ι1.0ζ)~ Ζ(^Ι5λ) Ρ\βκ = Ζ(^11.0λ)~ 2(^15J?) (23) Fi7 Scye Deep (Scye Depth): The vertical twist difference from the side neck point to the underarm point is the left and right neck point FholFh.or line, the left and right armpit points F2.olF2.〇r line, take the Z value of the midpoint of the two lines Poor; -^3.0£ +·^3·0/ί F2qL +F2〇r ^ 3 (twenty-four)? 18 thigh length (1'1^111^1^11): from the kneeling point to the knee width The vertical height difference is the z-value difference from the kneeling point to the knee width average height; A = Z (corpse 5.0) - muscle J + also) (25) 胯 high (internal leg length) (CotchHeight): from underarm Point to the vertical height difference of the heel 68 1315042 points, that is, the lower point Fs.q to the left and right heels Point fh〇l Fii.or the average height of the Z value gap; — Vdist (Fu (^, FS (i) + Vdist [FUQR, FSQ) (twenty-six) skill 0 shoulder height (Shoulder Height): from both shoulder points The average of the two vertical height differences to the heel point, that is, the average height of the two shoulder points F3 to the Z value of the average height of the two heels Fii.olFii.or; f2Q - thief ^ (^Ιι〇λ, 6〇 λ) + thief, skill. g) 2 (27) FZ1 NeckHeight: The vertical height difference from the side neck point to the heel point, that is, the average height of the two neck points Fu.olFuqr to the two heels Fii. OLFii.or average height Z-value difference; p' — Vdist (F-brittle, Fi30L) + Vd峨, F, 2 (28) Fa胯 waist (BodyRise): from the waist width level to the underarm point The vertical height difference, that is, the horizontal waist width F5 level height to the Z value of the underarm point F; F22 -z{Fs)-z{Fso) 2. The human body contour, the Z value is the largest point, the Xiang z value is not only - Point, take the second, the side of the projection amount (a) heart. 0 Head vertices (Side Head Point) The middle of all points is the head vertices; ^1.0 eQf, Sl0 = {Ρ.{χ, ζ) \ V /, ζ = Ζηαχ(Qs)} (2) S" side waist width (SideWaistBrea her): between the lower abdomen point and the middle point of the lower point, the shortest horizontal spacing of the front and back sides of the front and back sides, that is, the average degree of the mean and the lower limit of the FZ0LF2 Between the heights of the points, take the point group of the back wheel 1315042 with a z value between 1/2 and 3/4. If face right(), take the coordinate point of the smallest X value if it is not face right() , then take the maximum X value of the coordinate point, the horizontal distance from the curve of the precursor; Suppose Q„ear bachmisl = {P( | Ps 6 Qsb, (Ω4) + (Ωs)) &lt; ζ(Ρ{) &lt; (Ω ,) + 2^ (Ω,))} if face_right〇, χ(Ρα) = Μαχ^)), Ρ, e ^near backwaist if not face_right〇, x{Pa) = Minix^)), P{ e 5 , =HdistCli(Pa,Pb),wherez(Pb) = z(Pa),Pb ^Qsf _ (3)S2.Side Heel Point: In the calf to sole curve, the point of maximum curvature, ie The minimum Z value of the contour to the bottom 1/8 range of the maximum z-value of the contour, and the maximum BV value of the back body curve is obtained; * Suppose ΩΛβ0Γ JifeAee/ _ {/&gt; IP. g , (Ω^) &lt; z{ P.) &lt; (~ (Ω5) +—(Qs))j 8 8 s Let A: = 35 => Vi&gt; e Qeear_iiJeh Ed 3 520 = Max(BV(I))) (4) S2 height (HumanHeight): the height difference between the heel point to the head apex, that is, the distance from the heel S2.〇 to the top Sm at the Z value; =1 z(Sl 0 ) ~ z(S20) ® (5) & WaistHeight: The difference from the lateral waist width to the vertical height of the heel 'ie the horizontal waist width Si height to the heel point s2.0 z value difference; =Ζ〇ζ( 2020) (6) S4 side hip width (side Hip Breadth): below the horizontal waist width, above the squat point, the longest horizontal straight line distance between the curve of the front body and the curve of the back body 'that is the side waist width S1 to 胯The point group between the z-value heights of the next point, if face right (), take the coordinate point of the largest X value, if not face right (), take the coordinate point of the minimum X value, and the horizontal distance from the curve of the precursor; 1315042 0 gas says but = {乃丨乃e Ω^,z(8)&gt;z(乃)&gt; z(F50)} if face_right〇, x{Pa) = Max{x{P^ ?t eQneor jWeA(&gt; if Not face_right〇, x{Pa) = Min{x{P^ P, ^ S4 = Hdistni {Pa,Pb), where z(Pb) = z(Pa), Pb e Qs/ (seven) S5 leg width (Side Thigh Breadth): at the height of the neck, the horizontal width of the thighs is relative to the positive Calculate the horizontal width of the two contour points of the Fw and the height of the Z value; ^ Suppose {ρβ,ρ4 I χ{ρα) Φ x{pb)j Z(PJ = z(pb) = z(&lt;FSS )\Pa 6 Pb e Ωί4} => Ss=Hdista{Pa, Pb, (eight) & Knee Breadth: The side of the most prominent point of the knee 'water = wide distance' is relative to the front knee The width of the Ful'Fur average height is the same as the two contour points of the z-value height, and the horizontal width is calculated; Suppose |pe,Pft 14Pfl) ^ x{Pb), z{Pa) = z(Pft) = £(5sl) ±£(^ g ^ ^ ^ =>S6=Hdistn, Pa, Pb, (9) Side Mid Thigh Breadth: between the kneeling point and the side knee width • The horizontal width of the midpoint of the twist 'Complete the horizontal width of the two-wheeled corridor with the height of the leg F10 and the Z value in the front; S7 = Hdistcl(PaiPb) ▲e milk), shape milk) = z(Fm), Pflefv, p4ei ^ (10) S8 Min Min Calf Breadth: The minimum horizontal width between the knee width and the lowest point of the contour, that is, between the leg width S5 and the Z value of the heel S2_〇: Find the minimum horizontal distance of the contour; Suppose Qc〇// = {i,.,6 Ω /&gt;ζ(52〇) &lt; z(/,} &lt; z(5j)} 71 13 15042 Ss ^MHHdism^)) (11) S9 Calf Breadth: The maximum horizontal distance between the knee width and the minimum leg width, that is, the contour between the knee width S6 and the heel 38 Maximum horizontal width; Suppose = {/:. I e Ω5/, ζ(58) &lt; ζ{Ρ() &lt; Z{S6)} S9 =A/ax(/WiJi(Qca;/)) b)Sip Hip Height: from the lateral horizontal hip width to the heel The difference in vertical height, that is, the difference between the hip width S4 height and the heel point s2〇: S10=z(S4)-z(S20) (13) Su Tigh Length: From the kneeling point The difference in the vertical height to the knee width, that is, the difference between the kneeling point F5. 〇 to the knee width S6iZ value; 5n = z (F50) - z (S6) (fourteen) S12 胯 waist (Side Body Rise): by the side waist The vertical height difference from the wide height to the underarm point' is the Z-value spacing of the horizontal waist width Si height to the underarm point f5〇; S12 = z(St)-z(F5〇) (15) S13 Foot Length: The maximum foot length, that is, below the minimum leg width s8 height, summing the maximum horizontal width of the heel point S2. S; Suppose Ω / 〇ΰΙ = {P. I p. 6 Ω,,ζ^ίΩ,) &lt; z{Pt) &lt; z(58)} S9 =Max{Hdist{^lf〇〇t)) (16) SM side chest width (ChestDepth): 腋下点The height of the side of the chest is thick, that is, at the average height of the left and right squat points F2.0L, F2.0R, the horizontal distance between the front and rear body curves of the side profile; S14 = Hdist (Pa, Pb) 72 1315042 type method, Among them, the Γ: 1 to take the golden ratio of 0618: 丨 is better. 5. The method for constructing a three-dimensional digital human body model from a photo as described in claim 1, wherein in the hip-to-squat area, since the hip geometry of each person is not the same, the photo contour curve and The model model does not necessarily have the same length from the hip to the base of the thigh. If the χ value on the model is directly moved to the opposite position of the contour, even if the contour of the 3D side view is the same as the original photo. Contour, but the three-dimensional information distortion caused by the long thigh root will cause unreasonable deformation results in the lower abdomen. The solution is to establish the extension curve of the side abdomen contour and the extension curve of the thigh root upward. The practice of the line 拟合 fits two quadratic curves from the known contours as the extension line of the leg root and the abdomen, wherein the extension line of the lower abdomen is on the curve of the front half between the waist circumference and the underarm circumference. The maximum bending value of the point corpse extends down to the height of the side squat point, and the total length of the curve extending upward from the root of the leg is from the point to the curve on the front body. 111-fold between the different points of the high points of the curve length of the two extension lines determined can be considered as part of the contour of the curve, so that the extreme value on the model template shift therewith _ referred reasonable results recorded. Do not move to 6. As described in the fifth paragraph of the patent scope, the three-dimensional digital human body formula f&quot; from the photo, wherein the m value is better than U, this value is measured by a variety of different lengths. Good result. 7. As described in the third paragraph of the patent application scope, from the photo construction of the three-dimensional digital human body model method ^ in the shoulder circumference to the underarm area, to know the point movement of any point must be read 'The amount of the gift and the amount of rotation of the front and rear arm circumferences' can be known by the wheel in the direction of the movement of the middle tank. Therefore, the problem lies in the completion of the ship in the process of the ship. = The information on the front contour map can be deformed and the shoulder point can also be set as a known point, so any non-known 74 1315042 on the arm of the model model arm Ih ) Piih .+1)/2)|+|p(i_ )~p^t. +//+1)/2)|) Most j, outsourced 2D feature length = tin 1=0 (on tl Point o ' « _ table features around the small sections mainly through the three-dimensional method, the film and the side photos are read by the computer, the image obtained by the human body is processed manually, after circle selection background, profile Hf before ^ remove the noise point 're-search to obtain the image on the human body boundary ^ ^: Based on the basis: the model model, and reconstruct the body width two == count ε = \p^nti+x)\ / /(]pi.ti )-P((ti +ti+l )/2)|+|p(i. )_p((i. +ti+l)/2)\) Minimum Outsourcing 3D Feature Length =Sn )-作,)| The margin of each small segment around the feature on ι=0' W+1 table features small segments 77