Research And Application Of Median-type Filtering Algorithm For Resources Constrained System

The central task of electronic system is information processing. Digital filtering is a very important information processing technique. Resources constrained system has limited resources such as limited computing capability and memory size and so on. So the signal processing method of resources constrained system must meet the demand of simplicity and efficiency. In principle, median filtering is such a kind of filter that is suitable for resources constrained system. Many median-type filtering algorithms are proposed successively to improve some performances. They are widely used in image processing depend on the good performance of edge preserving and noise suppressing. They are also used in non-image processing because of the good overall performances, but for resource constrained system, the filtering process is redundant when the filtering window increasing. It is important to research and improve median-type filtering method to adjust the requirement of on-line non-image signal processing of resources constrained system.According to the requirement of signal processing of resources constrained system. This thesis references the existed median filtering algorithm, adopts the filtering structure of the median hybrid filters. Two algorithms separately improving the time and frequency domain are proposed. Using window decomposition and medians combination strategy, the proposed filtering algorithm has good comprehensive properties such as white-noise suppression, pulses suppression and simple filtering processing to meet the on-line signal processing requirement. Recursive structure is used to improve the frequency response of the algorithm. The filtering ability is simulated. To improve the frequency selective properties of the weighted median filters, a two-stage weighting algorithm is proposed. Using Particle Swarm Optimization algorithm, the weighting coefficients for the sub-window have high fitting degree. The simulation results show that the two-stage weighting filtering has a good bandpass characteristic. The computation complexity of the algorithms is analyzed.Wireless Sensor Network (WSN) is a typical resources constrained network system. The median-type filters are applied to the information processing of WSN in this paper. A fault-tolerant data aggregating algorithm is proposed which uses the proposed filtering algorithm and weighing method to realize fault-tolerant data aggregating of WSN under a necessary redundancy. The method has low power consumption and can meet the requirement of WSN application.This paper presents two kinds of median-type filters. The methods are universal signal processing ways, very adequate to resources constrained system. Median-type filter are expand to the application of data aggregating of WSN. The proposed methods are useful exploration and have good practicability in resources constrained system information processing.