摘要：AbstractInfrared image-based copper electrolysis short-circuit detection is a very difficult task because of compact distribution of electrodes and environment disturbance. The electrode hanger bars in the busbar area of the copper electrolysis cell are used as the direct diagnosis basis for short-circuit detection. To distinguish the hanger bars, which are small and dim targets with special shape, an improved DoG filter is proposed. First based on the prior shape information, a new target parameter model is proposed. Then the corresponding filter operator is used for DoG scale space constructing. The target model can more accurately matches with the filter operator and increase the detection accuracy. Most normal work hanger bars detected, according the detected hanger bar, short circuit ones can position indirectly. With this method, missing detection of short-circuit can be avoided.
关键词：KeywordsInfrared imagesshort-circuit detectioncopper electrolytictarget modelDoG scale space
