Electrical Resistance Tomography Edge-detecting Based On The Wavelet Transform

Electrical Resistance Tomography (ERT) is a kind of Process Tomography(PT) usingresistive sensor mechanism, which is suitable for detecting the parameters of the two-phaseflow. So far, it has been playing an important role in the industrial process due to itsadvantages of non-invasion, fast-responding, low cost, non-radiation, as well as flexiblemeasurement options.Edge detecting holds a special place in image processing and computer vision, which isone of the most important parts in primary visual processing. The quality of the detectingresult has a direct influence on the image compressing and computer vision, and it attachesgreat importance to the researches related. This paper, with the ERT images obtained fromlinear back projection (LBP) reconstruction algorithm, conducts the edge detection on theseimages using classical detecting operators, and introduces the wavelet transform afterconsidering the disadvantages of edge detecting methods. In addition, image denoising isperformed with different wavelet base functions, and the comparisons of their results withthose obtained with classical methods are made.To be specific, this paper involves the following aspects.1. Adopted the LBP algorithm to accomplish the work of image reconstruction,discussed the difference between the original and the reconstructed images and analyzed thelimitation of the algorithm.2. Programmed the five classical edge detecting operators: Roberts、Sobel、Prewitt、 LoG and Canny. Proposed an improved Sobel operator based on the limitation of thedetecting results, which could improve the precision of image edge detecting.3. Designed a MATLAB GUI panel. Threshold could be set using GUI tools in order toinspect the real time changes of images with the changes of thresholds.4. Introduced the technique of wavelet transform aiming to improve the imageresolution and precision, conducted deintegrate and denosing of noise image using differentwavelet base function, made comparation of different types of noise image using both hardand soft thresholds.5. Introduced the tool box of the wavelet GUI tool, through which one couldunderstand the powerfulness of wavelet analysis. Images being analyzed could be led intothe system and analyzed by different modules in the tool box.