Study On The Application Of Digital Image Measure Technique In Geotechnical Engineering

This paper conducts image measurement studies on national invention patents "Digital image measurement method and equipment of soil sample deformation in triaxial tests" (01113831.9), national science fund "Study on the improvement and application of the digital image technique for measurement of the sample deformation in triaxial test" (50527803), national science fund "Study on soil stress-strain features based on soil sample local deformation measurement in triaxial test" (50608014), "alphalt concrete deformation measurement" and "metal wire deformation measurement".Optical measurement regards images as methods or media by which the information is tested and transmitted. It extracts useful signal from images and processes images to obtain parameters. Optical measurement has non-contact, the whole visual field measurement, high precision and a high degree of automation features.Image processing both in theoretical research and in practical applications have made great progress in recent years. Especially with the application of computer technology, remote sensing technology, the development of digital communications, the popularity of computer network and the increase in microelectronic chip density, these factors have played key roles in promoting the digital image processing technology development.Sample images are obtained by optical sense and signs which are printed on sample are located with image processing methods. Location of these signs discovers deformation of sample. This paper focuses on how to relize measurement, accuracy, speed, stability and how to reconstruct whole deformation field from discrete points.(1) Line detection measurement of soil sample deformation is decribed in this paper.This method realizes soil sample deformation measurement with image processing technology first time in China. White lines are printed on the surface of the soil sample. Locations of the lines are obtained from the sample images. The height change between two lines is just deformation of the soil between lines.Deformation accuracy is improved compared with taditional measurement because one dimension subpixel algorithm is used.(2) Limitation of line detection resolution is discussed.If we don't consider the image noise, the maximum errorε_max= (1.0 - 0)/2 = 0.5 pixels.When the edge is an idea line, the minimum errorε_min =0. So we can infer that, (?)=(ε_max +ε_min/2 = 0.25 pixelsWhere,εis average error. In fact, object edge is usually rough and image noise always exists, so the better algorithm can only reach 0.2 pixels resolution. When the distance between two edges is measured, the resolution may be improved because that subtraction erases some errors while the improvement is limited.(3) Corner detection measurement of soil sample deformation is decribed in this paper. In order to improve accuracy, application of corner detection method in soil samplemeasurement is proposed first time. In the same time, this kind of method increases more measurement points than the line detection method. More points facilitate finite element analysis and contour analysis of the displacement and the strain field.Signs printed on the soil sample are changed in order to relize corner detection. White lines are replaced by white block. The four points of every white block will be located to calculate deformation of soil sample. Corner detection breaks through the accuray limitation of line detection. Standard deviation of measurement achieves 0.02 pixels. Time interval has been shortened from 2 seconds to 0.5 seconds.Isobar analysis of surface displacement field is discussed in this paper.(4) Signs-Structure method is proposed firstly in this paper in order to recover 3D displacement of soil sample based on only one image.The main idea is that the different distances of signs on the soil sample discover the 3D structure in the image.(5) Membrane penetration, end restraint and shear band are discussed with test data. Whole deformation is compared with local deformation with the test data which obtainedby image measurement. The result approves that two kinds of deformation are different. Traditional measurement uses the whole deformation as local deformation. It shows the limit of traditional measurement. The research shows that shear band place can be forecasted before shear band appears if the digital image measurement technology is used.(6) Application of digital image measurement technique in the alphalt concrete deformation is discussed in the paper.