CN106408598A - Three-dimensional portrait reconstruction printing device based on array lens - Google Patents

Abstract

The invention discloses a three-dimensional portrait reconstruction printing device based on array lenses. According to the three-dimensional portrait reconstruction printing device, a plurality of groups of calibration images are shot by the array lenses at first, the obtained images are calibrated, then a target object undergoes 360-degree all-dimensional shooting by means of the plurality of groups of array lenses so as to obtain a plurality of target images, the target images are subjected to three-dimensional matching to realize three-dimensional reconstruction, a three-dimensional model is drawn, and a stereoscopic model of the target object is printed by a 3D printing equipment. The three-dimensional portrait reconstruction printing device comprises a set of software system and hardware equipment, realizes the effect of carrying out real-time shooting and printing on a human body by means of the plurality of groups of array lenses, is high in realism degree of the printed portrait, and has good effect.

Description

A 3D portrait reconstruction printing device based on an array lens

technical field

The invention belongs to the technical fields of image processing and computer vision and computer graphics, and specifically includes a set of software and hardware systems, which can construct a three-dimensional model of a target object by performing three-dimensional reconstruction on multi-view stereo matching of multiple images, and can finally print it out.

Background technique

3D scene photography and 3D reconstruction are new hotspots in today's image processing technology, which integrates new methods and technologies in computer vision and computational photography such as light field imaging theory, 3D imaging, image-based 3D scenes, and super-resolution. In terms of algorithms for 3D reconstruction, traditional stereoscopic imaging systems use two cameras placed together to look at the object to be reconstructed in parallel. Conceptually, this method is similar to how humans calculate depth by superimposing images perceived by both eyes. If the distance and focal length of the two cameras are known, and the captured left and right images can be superimposed successfully, the depth information can be quickly obtained. Push it. This method relies on effective image pixel matching analysis, which is generally achieved using block comparison or epipolar geometry algorithms. Stereo vision using two cameras is also known as binocular vision, trinocular vision and other extensions using more cameras. The algorithm we use has been improved on the basis of the traditional stereo vision method, using a multi-camera stereo matching method on multiple lens arrays to achieve accurate 3D reconstruction.

At present, 3D printing technology and its related industries have achieved rapid development. With the diversified development of 3D printing materials and the innovation of printing technology, 3D printing not only shows extraordinary development potential in the traditional manufacturing industry, but also its charm is more It extends to many fields closely related to people's life, such as food manufacturing, luxury clothing, film and television media, and education. However, the 3D reconstruction and 3D printing technologies currently used in the market are still immature and the equipment is too expensive, which is not conducive to the popularization and application of new technologies. Therefore, we designed this system application integrating 3D reconstruction and 3D printing. to satisfy the market's needs.

Contents of the invention

The purpose of the present invention is to provide a new device for performing three-dimensional reconstruction and printing out a three-dimensional model of a portrait from multiple groups of array images obtained by an array camera.

The present invention adopts the lenses provided by Hikvision to form an array camera and a 3D printing device. The 3D reconstruction is realized through multi-eye stereo matching, and the 3D portrait model is finally printed.

Description of drawings

Fig. 1 flow chart of the present invention

Figure 2 Array lens used in the present invention

Figure 3 A set of calibration board diagrams

Figure 4 Partial images captured by the array camera from multiple angles

Figure 5 is selected as the center image of the reference image

Figure 6 software system interface

Figure 7 3D reconstructed and drawn three-dimensional model diagram

detailed description

In order to make the purpose, technical solution and advantages of the present invention clearer, the three-dimensional reconstruction and printing method of the device of the present invention will be further described in detail below in combination with specific examples and with reference to the accompanying drawings.

In the present invention, a test image is first obtained by shooting with an array lens, and the self-developed supporting system software is imported, and the camera is calibrated by Zhang Zhengyou's camera calibration method, that is, the internal and external parameters and distortion coefficients of the camera are calibrated. The center image is used as the reference image for multi-view stereo matching, and the depth map of the main objects in the scene is restored by the multi-view constraint hierarchical MRF algorithm for spatial surface reconstruction. 3D reconstruction. The flow chart of the device of the present invention is shown in Fig. 1 .

Step 1: Array Image Acquisition

Four sets of array images are captured by the same focal length array camera at one time, and each set of array images includes 9 pictures. In order to obtain the three-dimensional model of the object, the traditional technology needs to scan the object. There will be a certain time difference between different parts and different angles. There will be a certain error, but the array camera avoids this defect and has higher precision, which can make the 3D reconstruction result more accurate. FIG. 2 is an array lens used in the present invention, and FIG. 4 is a partial image captured by multiple array cameras from multiple angles.

Step 2: Camera Calibration

The array lens captures 20 sets of calibration plate images from different angles, and Figure 3 shows one set of calibration plate images. After acquiring the image, import the supporting system software for calibration (Fig. 6 is the system interface of the software). The camera is calibrated by the classic Zhang Zhengyou camera calibration method, and the internal and external parameters and distortion coefficients of the camera are obtained through OpenCV programming.

Step 3: Reconstruction of 3D scene by multi-eye stereo matching

The central image is used as the reference image for multi-view stereo matching. Figure 5 shows the central image selected as the reference image. The spatial surface is densely reconstructed by the multi-view constrained MRF algorithm. The depth layer interval is determined by the distance between two adjacent depths on the reference image The corresponding pixel position difference is determined. After the depth map is obtained, the camera parameters can be used to transform it into a spatial 3D point cloud to achieve 3D reconstruction.

Step 4: Reconstruct the surface of the 3D object

After the 3D reconstruction of the scene is realized, the 3D coordinates of any point in the image in the world coordinate system can be obtained through the projection relationship between the image and the 3D space, and the outline of the target object can be drawn using OpenGL programming, and color and texture details can be added. The final effect is shown in Figure 7.

Step 5: 3D Printing

Convert the reconstructed model into st1 format, and input it into a 3D printer for printing.

The above experimental verifications have described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be noted that the above descriptions are only individual experimental verifications of the present invention, and the present invention is not limited to the examples shown. Therefore, within the principles of the present invention, any modifications, equivalent replacements, improvements, etc., shall be included in the protection scope of the present invention.

Claims (4)

1. a set of equipment that the target objects such as human body can be shot and carried out with real time print, this equipment includes：

The array lens group that one group can be taken pictures simultaneously；

One 3D printing equipment；

Supporting software system, can demarcate to the image of acquisition of taking pictures, Stereo matching and three-dimensionalreconstruction, and controls 3D printing Machine carries out the printing of threedimensional model.

2. array lens group according to claim 1 is provided by Hangzhou Hikvision Digital Technology Co., Ltd, by 4 The array lens composition of group 3 × 3, once shoots and obtains 36 coloured images.

3. array camera according to claim 1 shoots and three-dimensionalreconstruction, and step includes：

1) array camera is demarcated；

2) multiple series of images that synchronization is shot carry out demarcating, distortion correction；

3) carry out the Stereo matchings that regard more rebuild three-dimensional scenic surface；

4) three-dimensional coordinate at image midpoint is obtained by image and three-dimensional projection relation；

5) threedimensional model of portrait is drawn out using OpenGL；

6) print portrait stereomodel.

4. the three-dimensional reconstruction method according to right 3 it is characterised in that multiple images obtaining are carried out regard Stereo matching, Realize the dense three-dimensional reconstruction of scene, obtain the detailed information such as the color and vein of target object simultaneously, and programmed by OpenGL Repaint out threedimensional model.This device should print portrait, can realize real-time three to the object of any appropriate size Dimension reconstructs and prints.