| Coarse aggregate is an important road material,and its quality directly affects the strength and stability of the road surface.Its shape characteristics are considered the most important features.However,the current control method for the shape and quality of coarse aggregates is still an empirical manual detection method,and there is a lack of evaluation standards for the overall shape and quality of aggregates.This article is based on digital image technology to collect video images of stacked aggregates on conveyor belts.Extract video frames,use digital image processing to obtain the two-dimensional contour shape of the surface particles of stacked aggregates,and use equivalent ellipses to fit the aggregate shape.Reduce fitting errors by repeatedly processing continuous video frame particle images.Based on the evaluation indicators of aggregate shape quality calculated from video frame images,a standard for evaluating the overall shape quality of aggregates is proposed.The main research content and conclusions of this article are as follows:(1)Extract the stacked aggregate video frame image,and conduct graying,histogram equalization,threshold segmentation,Gaussian high pass filtering and other processing on the image to preliminarily obtain the aggregate particle contour.Then perform morphological processing and watershed segmentation on the image to obtain a non adhesive aggregate particle image.Finally,the Graham convex hull algorithm is used to reconstruct the aggregate contour and achieve recognition of the surface particle contour of stacked aggregates.Using an equivalent ellipse to fit the shape of the aggregate,and using the long and short axes of the equivalent ellipse to fit the length and width of the aggregate,the extraction of particle size parameters on the surface of the stacked aggregate is achieved.(2)Adjacent video frames will overlap,and aggregate particles will be processed multiple times.Firstly,extract representative frames from stacked aggregate video images,which do not overlap and cover all aggregates,meaning that each aggregate is only calculated once.Then set constraints on the centroid coordinates and area of the aggregate particles on the representative frames to achieve tracking and recognition of the same aggregate particle on different video frames.And calculate the average size parameter of aggregate particles identified multiple times.The results show that after multiple repeated processing,the fitting error of size parameters can be reduced.(3)According to the aspect ratio,the aggregate shape is divided into square,oblong,and needle shaped shapes.The equivalent ellipse axial coefficient corresponds to the aspect ratio of aggregate particles.Firstly,the long and short axes of the equivalent ellipse of the representative frame aggregate particles are extracted,and the axial coefficient is calculated to obtain the recognition results of the aggregate shape image.Then,the representative frame aggregate particles are tracked and located,and the axial coefficients are calculated using the mean values of the long and short axes of the equivalent ellipse calculated multiple times to obtain the average video frame particle image recognition results.The results show that after multiple repeated processing,the accuracy of aggregate shape image recognition is improved.(4)According to the evaluation of aggregate shape quality,the standard deviation of form factor is proposed to grade aggregate.Based on the regulations on the content of needle and flake in the specification,the proportion of square,oblate,needle and flake aggregates and the form factor of aggregate particles are used.The aggregate is divided into three grades: excellent,good and poor.Through processing the continuous video frames,the mean value of aggregate particle form factor is obtained,and the standard deviation of aggregate overall form factor is calculated to evaluate the aggregate overall shape quality.(5)Develop a fast real-time detection and evaluation system through mixed programming of MATLAB and LabVIEW.Write an image processing program using MATLAB and embed the program into LabVIEW,which is directly connected to the camera to achieve fast real-time detection. |
PDF Full Text Request