| At present, the verification methods of mixing proportion for asphalt mixture at home and abroad mainly have centrifugal extraction, biological decomposition, nuclear method, combustion method and measure method of photoelectric instrument. These methods are very difficult to improve their accuracies during the concrete verification as their respective deficiencies. Therefore, on the basis of many related researches for reference at home and abroad and combining digital image technology, the dissertation takes the verification of mixing proportion for asphalt mixture as the research problem to develop the recognition research of mixing proportion for asphalt mixture based on the digital image processing technology. The researches mainly include 4 respects: the mixing proportion design and the image collection of asphalt mixture, the research of digital image processing technology, the asphalt-aggregate ratio recognition of asphalt mixture and the gradation recognition of asphalt mixture.(1) The Marshall tests have made at 5 different conditions of 4.0%, 4.5%, 5.0%, 5.5% and 6.0% of asphalt-aggregate ratios with the fine and dense graded type of asphalt mixtures AC-13C. The physical and mechanical indexes of samples were measured, and the optimum asphalt content was determined in accordance with the demands of specifications. Then the 3 samples of various asphalt-aggregate ratios were respectively cut as 2 pieces to collect the digital images with the horizontal separate method.(2) Based on the analysis of basic theories and related techniques of digital image processing, the collected digital images were carried on the comparisons of image restoration, image intensification and image division method. The results indicate that the primary RGB images are transformed into the gray images and the two-value images with the Lucy-Richardson algorithm restoration, the middle-value filter enhancement and the edge division of Canny operators, then the images will be clear in boundaries and easier to discriminate asphalt and aggregates.(3) The researches were made about the separation of asphalt and the calculation of asphalt-aggregate ratios in the gray images of asphalt mixture. From the gray histograms of asphalt mixture, the numbers of pixel for the wave peak of gray range asphalt distribution between 2.5σin left andσin right were extracted to calculate the numbers of pixel of aggregate and the recognition asphalt-aggregate ratio. The results show that the method of recognizing asphalt-aggregate ratio with the dynamic selection of gray range for asphalt can reflect the real asphalt-aggregate ratio in a better way. There is a good linear functional correlation between the real asphalt-aggregate ratios and the recognition asphalt-aggregate ratios. The linear fitting function was created as Pay = 1.6872Pax– 5.4606, and the correlation coefficient R2 is 0.9617.(4) The researches were made about the recognition of gradation of asphalt mixture. From the two-value images of asphalt mixture, the aggregate particle-size and the percentage of sieve residue over 2.36mm were recognized and calculated, and the recognition results were modified. The results show that the recognized gradations is close to the real gradations for each class of aggregates, both of them are good in correlations and smaller in dispersion degrees. The errors between modified recognized gradations and real gradations are smaller than 1%.The mass percentages of various aggregates above 2.36mm were obtained and we corrected the recognition results .The results indicate that the variance degrees of the recognition gradation data are very small, and the recognition gradations and the actual gradations have the good correlation.In brief, the dissertation has proposed a set of feasible method for obtaining the composition characteristics of asphalt mixture and established the recognition system for mixing proportion of asphalt mixture. The system can provide a scientific basis for the verification of asphalt mixture.
|
PDF Full Text Request