Partial Discharge Image Recognition Influenced by Fractal Image Compression

This paper presents a quadtree partitioning fractal image compression (QPFIC) method used for the partial discharge (PD) image remote recognition system. Self-similarity in PD images is the premise of fractal image compression and is described for the typical PD images acquired from defect model experiments in laboratory. Influences of fractal image compression on a group of PD image features are discussed. Fifty PD data samples are used to qualify the QPFIC to be used in remote PD pattern recognition. Analysis results show that the QPFIC method produces errors of the computational features. Such errors could not influence the PD image recognition results under the control of the PD image compression errors.