Research On RBF-based Image Reconstruction Algorithm For Electrical Capacitance Tomography System

Electrical capacitance tomography technique(ECT) is a process based on capacitive sensing mechanism of imaging technology.It is the mainstream of development and research focus in the current two-phase testing in the field of settlement of industrial process tomography technology and has great potential and broad industrial development prospects.It is the current process tomography technology mainstream and research focus. However, in the practice there are some problems and specific difficulties, the current research is focused in the laboratory research stage, there is a great distance to the large-scale industrialization, so it has to be done the further study.Reconstruction is the key to Capacitance tomography system,so the reconstruction algorithm of image is directly related to the realization of image reconstruction, and also is an important factor in the quality of image reconstruction. In this paper, the 12-electrode capacitance tomography is a hardware implementation for radial basis function(RBF) algorithm of image reconstruction design and hardware description language(VHDL) in the field programmable gate array(FPGA) on implementation.In compare with software implementation methods, it can effectively shorten the running time and maintain high classification accuracy. The study we have done is as followed:First of all, discuss the composition and structural characteristics of ECT deeply, analyze the operating principle of ECT technology, the existing typical image reconstruction algorithms is studied, analyze the advantages and disadvantages of each algorithm to obtain a better algorithm in order to inherit the advantages and make up for shortcomings. Based on radial basis function neural network image reconstruction algorithm is proposed, use the improved K-means algorithm to train the hidden layer's centers and widths of radial basis function, use the Tikhnov regularization method to determine the weights of network connection, establish a capacitance measurements and imaging permittivity distribution of the nonlinear mapping region, and this method improve the image quality.Secondly, use hardware description language to realize RBF-based image reconstruction algorithm on FPGA, this algorithm uses serial computing-parallel transfer mode, and describes the hardware system and the overall structure of the workflow.