Research On Data Compression Method Of Massive Magnetic Flux Leakage Detection Data Of Surface Defects On Steel Plate Based On Improved SPIHT Algorithm

With the acceleration of economic construction, steel plates are widely used in the field of automobile manufacturing, mechanical manufacturing, aerospace, petrochemical industry, shipbuilding and so on. The amount of surface defects is one of the most important indexes in judging the quality of a steel plate. Surface defects on steel plates not only do harm to rolling equipments, but shorten the life of steel products, even leading to industrial accidents. The detection of surface defects on steel plates is recognized as an effective method to prevent the negative occurrence mentioned above. Because of its superior performance, magnetite flux leakage detection is more and more widely applied in the detection of surface defects of steel plates. However, with the increasing of steel plate output and the improvement of measuring accuracy, the data acquired by the detector is very huge, which brings great difficulty to the procession and transmission of data. In order to improve the performance of detectors, it's necessary to work out a feasible data compression method to reduce the massive data. Aiming at the compression of the massive data acquired by detectors, the following work has been done in this thesis:First, in order to compress the huge amount of data effectively, the features of magnetic flux leakage data have been analyzed. And then some common data compression methods have been introduced. The simulation results of these common data compression methods, with the input of magnetite flux leakage data of surface defects on steel plates, have been given to analyze and compare their compression performance. According to the simulation results, a preliminary scheme to compress magnetite flux leakage data of surface defects on steel plates has been proposed.Second, two preprocessing methods have been discussed and the comparison of the performance of these two methods has shown that direct translation is better than differential pulse code modulation. Then set partitioning in hierarchical tree algorithm (SPIHT) has been studied. Simulation based on SPIHT has been illustrated. And performances of different wavelet bases, wavelet decomposition levels and SPIHT encoding levels have also been given.Finally, aiming at covering the shortage of the classic SPIHT algorithm, the optimization has been raised in the aspects of wavelet transform, zero-tree structure and storage of important massage. The comparison of simulation results between classic SPIHT and improved SPIHT algorithm has shown that the improved algorithm is with better performance in reconstruction error, which has proved the feasibility and efficiency of improved SPIHT algorithm.