Image Reconstruction Algorithm Based On SVM For Electrical Capacitance Tomography System

Electrical capacitance tomography(Electrical Capacitance Tomography, ECT)is a process tomography technique developed early, it is non-invasive, non-contact,low cost. At present, the main problems of ECT technology in the field of sensitive technology in soft fields, resulting in the process of image reconstruction and the imaging accuracy is not high, and can not meet the requirements of speed. Therefore,image reconstruction has become important link between with ECT technology.Support vector machines(SVM) is a kind of algorithm is widely used in the field of machine learning, which has good generalization ability, especially in solving nonlinear problems. As a result, the application of SVM in image reconstruction of ECT As a result, the application of SVM in image reconstruction of ECT technology provides the basis.Firstly, the overall structure of the ECT system and the basic principle in detail,respectively on the capacitive sensor, ECT system in data acquisition system and image reconstruction are described. This paper according to the analysis of 12 electrode ECT systems was studied, and the sensitive field model of 12 electrode capacitance sensors is established using the finite element software ANSYS. This model of area elements, the structure of the model has a symmetrical characteristic,and the application characteristics can be simplified to the measured data set.The in view of the application of SVM in ECT image reconstruction is analyzed,and found that SVM algorithm to reconstruct images of large scale and high dimension of sample data processing performance is not ideal, leading to the imaging precision is not high, the imaging time is relatively long. Based on this the paper proposes a data reduction operator based on the features of the cluster center(referred to as FRCO) method combined with SVM method. This method firstly usedthe FRCO method to reduce the dimension of data sets to simplify data characteristics, required to achieve the SVM algorithm suitable for image reconstruction. Experiments show that by combining FRCO and SVM used in enhanced image reconstruction, especially in imaging speed than using only the SVM image reconstruction has been significantly improved.