Abstract: The image produced by the reconstruction algorithm for electrical capacitance tomography can be inevitably affected by the outside disturbance and it is destructed in the process of obtaining and transmission. So, the images are affected by noise pollution and the feature extraction and image recognition of subsequent processing of the images are also influenced. In order to solve this problem, a noise reducing algorithm for Electrical Capacitance Tomography (ECT) is presented. It is based on the analysis of the basic principle of electrical capacitance tomography and Pulse Coupled Neural Network (PCNN). The time matrix is the mapping from spatial image information to time information generated by PCNN and the time matrix contains useful information related to spatial information in image processing. The calculation steps for noise reducing are deduced. The size of filter window and the number of filtering can be automatically selected according to the noise intensity. Experiment and simulation results indicate that through analyzing and processing the PCNN time matrix, the image which is polluted by impulsive noise can be filtered efficiently. The effect of reducing impulsive noise of images is significantly better than the median filter, mean filter and wiener filter. This algorithm presents higher peak signal-to-noise and it has better capability to reduce noise. It can also well protect edges and details of images and it is more adaptive.