TWI285843B - System and method of generating a virtual city model - Google Patents

Abstract

This invention relates to a system and method for manufacturing a virtual city model. The method employs a window graphic interface to build an operating platform, including steps of providing construction information, computing image preview, resolving visual azimuth, determining occluded areas, dealing with situations of being not occluded and partially occluded. This invention can generate large amount of texture images of curtain walls for constructions in a 3-D virtual city with the use of effective and quick visual resolving azimuth parameters external to the camera. It can also automatically compensate textures on self-occluded areas of the constructions in response to occurrence of self-occlusion phenomenon on the constructions, so as to achieve excellent image quality and quick and effective generation of texture images at the same time.

Description

1285843 Said,·.转克': 九, invention description: [Technical field of invention] The present invention relates to a method for making a immersive city model, especially a building texture in a main and three-dimensional imaginary urban landscape browsing Image. The technology of production. [Prior Art] Because the city is the most important base for human living environment, the urban environment is related to the quality of human life. The dynamic changes of buildings, roads, trees, various vehicles, and human beings are directly Or indirect interaction. Therefore, the production of 3D immersive city models is becoming more and more important because of its wide range of applications, such as disaster prevention and reconstruction, city navigation, real estate sales, education, historical digital collection, environmental impact assessment, e-commerce, mobile communication base. Simulation and planning of the location of the station. Press, the commonly used techniques are as follows: 1. As published by Haala, N., 2004., “On the Refinment of Urban Models by Terrestrial Data Collection”, Vol. 35, Part·B3, Istanbul, Turkey, pp. 564 -569. It is a GPS+ push-broom digital camera (parallel + perspective projection) orientation and geometry processing system. The house model is a real height polyhedron, and the number of processed walls is processed together with all visible surfaces; The disadvantage is that the equipment required is expensive, the shooting position must be at an appropriate high point, the post-processing workload is large, and any shading problems are dealt with. 1285843

.3⁄4 • Hemp bundle 2004. Issued 2. For example, UK / Iran Varshosaz, M., Table “ Occlusion-Free 3D Realistic Modelling of Buildings in Urban Areas" , IAPRS, Vol.35, Part. B4,

Istanbul, Turkey, pp· 437-442·It is a special equipment such as GPS+theodolite. It uses multiple simultaneous photo and geometric processing systems. The house model is a real high polyhedron, and the processed wall texture is used at one time. Image mosaic method, and one wall is processed at a time, and the shading processing is a method of compensating each other by using multi-view images; and the disadvantage of this method is that special equipment is required, the price is high, and the post-processing is time-consuming and labor-intensive. 3.·For example, “National Kada, M·, 2004· published” Hardware-Based

Texture Extraction for Building Facades" , IAPRS,

Vol. 35, Part B4, Istanbul, Turkey, pp. 420-425. It is a geometric processing system using planar perspective projection (eight-parameter conversion), the house is molded into a prism or a polyhedron, and the treated wall The number is one-face wall, and the shading treatment is a method of using multi-view image mosaic and post-processing repair; and the disadvantage of this method is that for complex houses, more labor is required than the present invention. Selecting a control point and using only planar projection conversion will not work for images captured by a wide-angle camera, resulting in a lower texture quality and a higher post-processing workload. Therefore, the above three methods are not able to quickly and effectively produce building texture images to suit the needs of actual use. SUMMARY OF THE INVENTION The main object of the present invention is to assist the 1285843 operator in solving the processing bottleneck of the camera external orientation solution by visualizing the interface. Another object of the present invention is to provide specular reflection _ because (4) (four) self-shadowing caused by insufficient texture ^ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The number of scene photos to reduce the workload and increase efficiency. In order to achieve the above purpose, the present invention is a method for fabricating a immersive city model and a method, which is constructed by an interface, an operation platform, a main t building data supply, an image preview calculation, a visual green solution, and a delivery site. Judgment, treatment without masking, and processing during partial masking: 'When performing, provide the real height model of the building's appearance and the scene = the appearance of the building's texture image'. Simulate the building in a three-dimensional view A scene that approximates the scene photo can be automatically or manually selected from the visible surface of the field where the building is not completely obscured, and the three-dimensional geographic coordinates of the roof angle and the foot of the visible surface are used as ground (four) points, and aerial survey technology is used. The single resection method calculates the initial external orientation parameter when the camera is photographed; the back projection calculates the corresponding image coordinates of the four corners of the visible surface, and connects the four corners to form a frame of the building model wall surface. When the operator dynamically adjusts the image control point, the camera's external orientation parameters can be solved immediately, and the back projection nesting of the building model frame can be performed, and then (4) (4) Whether the object model frame is positively aligned with the wall periphery of the building in the photo The visualization method completes the solution of the outer orientation of the camera; judges the presence or absence of the shadow phenomenon and calculates the position of the self-shadowing; in the absence of the shadowing phenomenon, the three-dimensional space coordinates of the equally spaced grid on the wall are back-projected 95. H4 1285843 calculates the image coordinates of the image of the appearance of the building in the field, and then obtains the complete wall texture image by interpolation; in the case of partial shadowing, the texture image of the unobstructed area is first generated and judged along the periphery of the texture image. The number of pixels without texture information, and the ratio of the number of pixels relative to the perimeter of the wall, defines the perimeter that occupies the largest masking ratio as the main. To mask the edge, use the main shadowing edge on the adjacent two perimeters. The maximum value of the number of masked pixels is used as the width of the specular reflection and the position of the mirror, and the shadow is automatically compensated by the specular reflection. The texture information, in order to get the # whole wall texture image. [Embodiment] Please refer to "Figure 1", as shown in the figure: The present invention is a method for fabricating a pseudo-real city model, which is constructed by a visual interface of a window and a drawing application software design, thereby constructing an operation platform as an operator. The software development interface with the computer, and the software development platform used is Microsoft Visual Studio . NET. The main drawing library is the OpenGL utility library, which provides 3D graphics acceleration for most image interface cards. The system mainly includes steps of building material supply 1, image preview calculation 2, visual orientation calculation 3, masking judgment 4, processing without masking 5, and processing 6 when partially masking. If so, a new method of making a immersive city model is constructed by the above arrangement. Please refer to "Fig. 2~1 3", as shown in the figures: When the present invention is used in the above-mentioned steps, its operation is as follows: a. Building material supply 1: Provide a real height model of the building appearance And on-site building appearance texture image data, the appearance of the building is 1284843

, 隹 building model 1 corner of the three-dimensional geographic seat The true height model is a j, , , , , 隹 标 ’ as a ground control point 盥 仏 仏 AL...1

1, eight records the roof of the building and the corners of the foot of the 11 〇 contiguous correction, so in the relevant calculations can use its lens distortion parameters while correcting its geometric deformation 'to facilitate the application of wide-angle cameras. b. Image preview calculation 2: Simulate the building model in a three-dimensional view. 1 1 A scene that approximates the live photo (as shown in Figure 5), which can be automatically or manually selected for visually obscured buildings in this scene. Face, and the three-dimensional geographic coordinates of the roof angle and the foot of the visible surface are used as ground control points. As shown in the (E, N, H) data in the lower left window 21 of Figure 6, the operator can look at the building. In the texture image, click the visible roof corner and the foot of the house as the image control point coordinates, as shown in the Line (Sample) data in the lower left window of Figure 6. When four or more evenly distributed control point pairs are selected, the present invention can use the single resection method in the aerial survey technology to instantly calculate the initial external orientation parameters of the camera, including the position (χ, γ, ζ). ) and posture (〇, >, six coefficients, as shown in the bottom right window 2 of Figure 6 (Χ, Υ, Ζ) and (ω, 必, κ) data. \ , c• visual orientation Solve 3: Because the source of the ground control point, the selection of the image control point and the lens distortion parameters have certain errors, the initial external orientation parameter obtained by the above method must also have an error. 2 Solve this dilemma, The present invention (4) calculates the image coordinates corresponding to the four corners of all visible faces by using the initial external azimuth parameter and the collinear conditional formula 'the back projection in the aerial survey technology, and connects the four (four) falling links to form the building model ^1 wall surface 3 1 The frame; t operator dynamically adjusts the image control point = can solve the camera's external azimuth parameters, and carry out the back projection nesting of the building model frame. Then judge the building model frame with the human eye: the wall of the building in the piece Surrounding The correct fit. As shown in Figure 6 on the upper right, if the orientation parameter of the camera is not correct, the money of the (4) model frame after the back projection cannot be correctly matched with the shape of the building. When the wall texture image is generated, the result must not be accepted. For example, the texture of the wall surface 3 1Jl in Fig. 7 can still see that the image of the sky is 'incorrect result; and when the building is reversed, the model 1 1 When the frame and the shape of the building are correctly assembled, the correct external orientation parameter solution is reflected. The resulting wall texture image will be as shown in Figure 8, except for the trees on the wall 3! 3 2 non-self-shadowing texture can not be removed, the rest are the same wall surface 3 1 texture image. With this visual solution process, in addition to the effective and correct camera orientation parameters can be obtained, and can maintain the most Good texture image geometry quality d. Obscuration judgment 4: Using the above camera external orientation solution results and wall geometry model, you can judge the presence and absence of self-shadowing, and calculate the position of self-shadowing. The figure shows that the highest wall 3 1 on the front has a wall covering the left side of the work. When the camera is obtained outside the field 1284543 : t • Jane _'丨:. After the static bit parameter, the 3D geography of the visible surface is used. Coordinates, back projection to image space _, and use the distance from each wall to the camera position to determine the shadow relationship between the walls, as well as the shielding position. See Figures 9a, 9b, which is from the sky overlooking the building A schematic diagram of the position of the object model 1 1 and the camera 7, wherein the camera of the 93rd view is centered at P1, the walls A, B and the walls C and D are respectively projected to the image plane 71 at positions a, b and c, respectively. , d, at this time a ' b and c, d do not overlap each other, so it can be known that there is no shadowing between the two. The perspective center of the camera in Figure 9b is P2, the same wall A, B and wall C, and the position of the image plane 71 are a, b and c, d, respectively, a, b and c, d There is overlap between each other, and since the distance between B and P2 is longer than the distance between c and P2, it can be judged that the wall c and D will be shielded to the wall surface A, B 〇" e· without shading Treatment 5: If a wall is not obscured, the collinear condition can be used to calculate the three-dimensional coordinates of the _ equally spaced grid on the wall, corresponding to the appearance of the building. The image coordinates on the image are then interpolated to obtain a gray image of the grid to obtain a complete wall texture image, as shown in Fig. ii, the texture image of the ga wall c and d. Treatment of the bite I part Λ 6 6 · · The invention assumes that most of the wall surface: the principle of repeatability or uniformity, so if a wall has a style, like the hair = when First, use the collinear condition: 'calculate the three-dimensional mark of the equidistant grid on the wall by the back projection method' corresponding to the image of the building's appearance texture image taken on site 1 12285843

The texture image of the current f is shaded. As shown in Fig. 10, the wall A, B, ', and ~ on the right side of the wall are not covered by the wall faces C and D. Then, according to the "texture image with masking phenomenon", the number of pixels without texture information is judged along the periphery of the image, compared with the number of pixels around the wall, and the periphery of the largest masking ratio is defined as The main shielding edge 6 1 , such as the main shielding edge 6 1 of the 1 2 a diagram, has a maximum shielding ratio of 1%. Then, use in the main

The maximum value of the number of masked pixels on the two adjacent sides of the shadow side is taken as the width of the specular reflection and the mirror position. For example, in the second graph, W1 is larger than W2, so wi is used as the width of the specular reflection, and a is b is used as the mirror position, and the texture of the dotted frame c in the figure is the texture information of the mirror reflection compensation shadow, and finally the complete texture image can be obtained, as shown in Fig. 2b.

The texture of the exterior wall of the building in the 3D symmetry city model is made by the system and method. The geometric model of the terrain is a 20-meter grid numerical terrain model, and the aerial photograph is used as the material. Applying makes the entire landscape simulation closer to the real world. In summary, the method for making a pseudo-real city model in the present invention quickly and effectively solves the camera's external orientation parameters in a visual manner, and automatically compensates for the texture of the building's shelter when the building partially self-shadows. At the same time, the purpose of satisfying the image quality and the rapid production of the texture image can effectively improve the practical disadvantages of the conventional use and the stability and the stability of the invention, thereby making the invention more progressive, more practical, and more suitable for 13 1285843 j % if The rash replacement negative / / user needs, has indeed met the requirements of the invention patent, and filed a patent application according to law. The above is only the preferred embodiment of the present invention, and the scope of the present invention is limited to the scope of the present invention; The simple equivalent changes and the decoration are still within the scope of the patent of the present invention. 1285843 Sheep Replacement Page"""" .11111•• Let one.'—一一 _二”"^ 0^^immk\u............... ........... 〆 [Simplified description of the drawings] Fig. 1 is a flow chart showing the steps of the steps of the present invention. Fig. 2 is a flow chart showing the flow of the present invention. The schematic diagram of the real high polyhedral three-dimensional house model used in the present invention. Fig. 4 is a schematic diagram of the appearance image of the building used in the present invention. FIG. 5 is a three-dimensional browsing method for simulating a three-dimensional house model. Fig. 6 is a schematic diagram of visually calculating the orientation parameter of the camera by clicking on the image control point of the present invention. Fig. 7 is a schematic diagram of the texture image generated by the incorrect orientation parameter. Schematic diagram of the texture image generated by the correct orientation parameter. Figures 9a and 9b are schematic diagrams for judging the mutual obstruction of the wall surface by the present invention. _ Figure 10 is a schematic diagram of the complete texture image produced by the non-shadowing phenomenon. Figure, is a schematic diagram of the texture image with partial self-shadowing The 1 2 a is a schematic diagram of the texture of the specular reflection compensation mask of the present invention. The 1 2b diagram is a schematic diagram of the texture image after the mirror reflection compensation of the present invention. The 13th graph is a schematic diagram of the 3D pseudo-real city model produced by the present invention. 15 1285843 [Component label comparison] = Building data supply 1 Building model 1 1 Building 1 2 • Image preview calculation 2, lower left window 2 1 lower right window 2 2 upper right window 2 3 #Visual orientation solving 3 wall surface 3 1 Trees 3 2 Masking judgment 4 Processing without shading 5 Processing when partially obscuring 6 Main shading edge 6 1 Camera 7 _ Image plane 71 Camera perspective center PI, P2

Claims (1)

1285843 / - '-1 $Please ask the members to express the 9th year of the year, the date of the month, the replacement of the page, the revision of the page, the presence or absence of the original specification, or the scope of the patent application: +. A method for fabricating a immersive city model is to construct an operation platform by using a window drawing interface, which at least comprises the following steps: a. Building data providing: providing a real height model of the building appearance and the appearance texture image data of the building; b Image preview calculation: simulates the building model in a three-dimensional browsing manner to achieve a scene similar to the scene photo, and can automatically or manually select the visible surface of the scene in which the building is not completely obscured, and the roof corner and the foot of the visible surface The 3D geographic coordinates are used as the ground control points, and the roof corners and the roofs visible on the exterior texture image of the building are manually selected as the image control point coordinates, thereby calculating the initial external orientation parameters when the camera is photographed; c. Visual orientation solving: The back projection calculates the image coordinates corresponding to the four corners of the visible surface, and connects the four corners to form a frame of the wall of the building model. When the operator dynamically adjusts the image control point, the camera's external orientation parameters can be solved instantly, and the back projection nesting of the building model frame can be performed, and then the building model frame is correctly aligned with the wall perimeter of the building in the photo; d. Judging the shadowing position··Using the above-mentioned external orientation solution result and the wall geometry model to judge the presence or absence of self-shadowing and calculate the position of self-shadowing; e. Processing without masking··Responsive calculation of wall The three-dimensional space coordinates of the equal-pitch grid on the surface correspond to the image coordinates on the appearance texture image of the building photographed on the spot, and then the interpolation process is completed to complete the processing of the wall surface f· partial shading. The method of e generates a wall texture image of the unobscured portion, and then judges the number of pixels without texture information along the periphery of the texture image, and defines the ratio of the largest masking ratio after the ratio with respect to the number of pixels around the wall. As the main shielding edge, the maximum value of the number of masking pixels on the two adjacent sides of the main shielding edge is used as the specular reflection And mirror position of 'reflection and specular manner automatically compensates for the texture information of the shelter, so as to obtain a complete image of the wall texture. • The method for producing a immersive city model as described in claim 1 'where the true height model of the building's appearance is a true height polyhedral three-dimensional building model that records the roof of the building and the corners of the house. The 3D geographic coordinates serve as a source of information for ground control points and building shape frames. • The method of producing a immersive city model as described in claim 1, wherein the on-site texture image of the exterior of the building is a digital image of the appearance of the building, as texture information and point-selection image control The source of the point. • The method for producing a immersive city model as described in claim 1 of the patent scope, wherein the camera external orientation parameter solving system utilizes a single resection method in the aerial survey technology for immediate calculation. • The method for producing a immersive city model as described in claim 1 of the patent scope, wherein the processing without occlusion uses a collinear conditional expression to calculate the three-dimensional space of the equally spaced grid on the wall by back projection. Block 1284543 Hi repair (5) is replacing the page mark, corresponding to the image coordinates on the appearance texture image of the on-site building, and then obtaining the gray value of each grid by interpolation to obtain the complete wall texture image. 6 · The method for producing a immersive city model as described in claim 1, wherein the processing of the partial occlusion is performed after the unmasked portion first uses the method of the fifth application of the patent application to generate a texture image. The texture information of the shadow is compensated by specular reflection to obtain a complete wall texture image. 1285843 nm!, ...r' VII. Designation of representative drawings: (1) The representative representative of the case is: (1). (2) Brief description of the symbol of the representative map: Building data supply 1 Image preview calculation 2 Visual orientation solution 3 Shadowing judgment 4 Processing without masking 5 Processing during partial masking 6 5