Research Of Segmentation And Statistics Of Asphalt Mixture Based On Image Procession

The asphalt pavement is the main pavement of senior and times senior road in China, the asphalt mixture is a major component of the asphalt pavement, it’s quality directly affect the service life of the road.Asphalt mixture testing and statistical techniques have important significance in the research of asphalt pavement. The asphalt mixture based on the CT scan have less noise, high definition, clear grain boundaries, it is relatively easy to managed, however, the need for specialized equipment,the cost is relatively high, dimensional asphalt mixture based on the camera, there are many unfavorable factors, segmentation and statistical results are not satisfactory, can not be effectively separated asphalt mixture there is a connection particles. This thesis detect more complex camera asphalt mixture, segment particles in the image, the corresponding statistical results given by the segmentation results.First,this thesis analyzes a more complex profile image of the asphalt mixture shot by digital camera, due to many disadvantages of the original image,thus image pre-processing is used, then Gaussian noise is removed from the image by transforming it into the gray-scale,removing the uneven illumination due to the external environment, sharpening the image fuzzy edge information, segmenting the gray image based on OTSU threshold segmentation method.Second, removing the wave noising, detecting the edge of the asphalt mixture image based on threshold segmentation method, combining the two binary images.Analyzing the binary image existed connection,Firstly, in order to maintain the shape of the aggregates, the image is carried out small-scale morphological operations for disconnecting the small connection, then proposing a iterative segmentation algorithm based on relationship between the concave point and the edge to remove the big connection. Firstly, according to the big connection is always concave characteristics,using the relationship between the concave point and the edge to do iterative segmentation.Analyzing the segmented binary image of the asphalt mixture, First determine the statistical methods of the aggregate particles, on this basis, give the overall geometric features and shape features of the image, draw the grading curve of asphalt mixture.The experimental results show that the asphalt mixture initial two segmentation can be a binary image of the glue regional and aggregate particles, the separation of connected particles able to largely maintain the number of small particles and large particles will not have too much deformation, only a small portion of particles there will be over-segmentation and under-segmentation phenomenon. The statistical results can accurately calculate the parameters of the asphalt mixture.