The ET Image Quality Evaluation And Analysis Of Algorithms Based On The Method Of Model Group

Electrical Tomography (ET), which belongs to the process tomography, is developed as an independent technology since the beginning of 1980s. ET technology includes many forms, but they both have the same features, such as high imaging speed, no radiation and non-invasive. With the advantages of simple structure and low cost, ET is really suitable for the applications in the biomedical or industrial field, and it is developing very fast.Though great progress has been made for ET technology, its development still has some constraints due to the quality of its reconstruction images is infected by the "soft-field" character and the under-determined and ill-conditioned property of the inverse problem. In order to obtain higher quality reconstructed images, many researchers paid attentions to the study of the reconstruction algorithms. Many works have been done by the researchers to evaluate the property of algorithms objectively with different kinds of evaluate methods. Compared with other methods, the method of model-group has some special property and it can evaluate the tracking ability of the algorithms to the objects’ changing, which is a new direction for evaluation of the algorithms. With this method, different kinds of model groups have been established to evaluate some classic algorithms, the tracking ability of these algorithms have been analyzed. The research process and results are worth learning for the evaluation of ET reconstruction algorithms. The main work is as follows.1. The modeling of ET, the achieve of algorithms and the image reconstructionBased on the understanding of ET theory, several model groups are established by the finite element analysis software COMSOL, some widely used algorithms are programmed by Matlab, the images are finally reconstructed by these algorithms and solutions from COMSOL.2. The design of model groups and evaluation parametersThere are four kinds of model groups in three classes have been designed. Different parameters are set to evaluate the tracking ability of algorithms to the object’s changes of size, shape, distance and electrical character for each group. Detailed analysis has been done according to the parameters’ results. The properties of algorithms are evaluated more comprehensive through the experiment. 3. The design of software for ET image evaluationBased on the method of model-group, ET image evaluation software is developed in environment of Visual Basic.NET and SQL Server. The system has a friendly interface, easy and standard operation method. It can deal with the reconstruction images in 24 BMP format and calculate the parameters successful. The software meets the need of evaluation and suitable for practical use.