Research On 3D Vision Structured Light Coding And Decoding Technology

Three-dimensional reconstruction is one of the most popular topics of computer vision,and has been widely used in industrial detection,object recognition,robot navigation and other fields.Structured light technology and binocular vision technology have been widely used in the field of three-dimensional reconstruction.The principle of binocular vision is relatively simple and easy to operate and has been widely used in three-dimensional visual inspection.However,it is difficult to obtain depth information in scenes with less texture or without texture,and the computational cost of matching is very costly.Structured light measurement technology can extract the image features accurately,and the accuracy is relatively high,but the process of projector calibration and non-linear correction is rather complex.According to the advantages of binocular vision and structured light technology,this paper combine binocular vision with structured light technology,and put forward a scheme of binocular structured light measurement system.In this system,the focus of this paper is the coding and decoding of structured light.In order to solve the problem that the surface of the measured object has no texture or less texture reconstruction and the precision is not high,we can increase the texture of the surface of the measured object by projecting the structured light.Then according to these texture features,we can decode it.In the process of structured light decoding,when we extract the color stripe centerline,the accuracy of system reconstruction is improved by introducing sub-pixels.The surface color of the object and ambient light interference to measurement system is solved by using color clustering algorithm.The main contents and innovations of this paper are as follows:1.We put forward a set of binocular structured light measurement system and compare with the traditional binocular vision measurement system and the traditional structured light measurement system to solve the problem of surface without texture or less texture that is difficult to reconstruction and the reconstruction accuracy is not high,as the binocular vision principle is simple and easy to achieve and structured light technology has high precision and is able to increase the texture of the measured object artificially.2.The binocular structured light measurement hardware system and software system are designed and a hardware system platform is built.A binocular structured light coding,binocular structured light decoding,three-dimensional reconstruction and other software function modules and human-computer interaction interface are developed.3.In this paper,we study the coding method and principle of structured light.According to the mathematical model of De Bruijn sequence,a modified non-repetitive De Bruijn color stripe coding is proposed to solve the problem that traditional De Bruijn color stripes may cause mis-matching problems.According to the principle of color camera and the advantages of gray camera,a method of obtaining color image through gray camera is proposed.4.In the process of structured light decoding,to solve the problem of serious impact on object surface color and ambient light,we use a color clustering algorithm based on HSI space to complete the color separation of colored stripes.The Canny operator is used to detect the edge of the colored stripe,and then the pixel centerline of the colored stripes is extracted by the improved geometric center method.The algorithm can eliminate the incomplete edge pair.In order to further improve the system accuracy,on the basis of this,we use adaptive threshold gradient gravity method to extract the color stripe sub-pixel center line.Experiments show that the introduction of sub-pixel centerline decoding of color structured light can greatly improve the accuracy of system reconstruction.