Abstract:

Given that the furnace flame noise image in coal power plants features mixed impulse noise and Gaussian noise, we propose a denoising method for furnace flame based on the median filter and wavelet transform. Firstly, we remove the impulsive noise from the furnace flame by the adaptive weight median filtering method, and then decompose the denoised image using the wavelet transform. And then different filtering methods are used for different sub-bands after the decomposition. Experimental results show that the proposed method can effectively remove the noise and improve the quality of the flame image, and it outperforms the traditional denoising method.

Key words: furnace flame, median filter, wavelet transform