Abstract: The denoising methods based on wavelet transform have been widely applied to eliminate the noise in the chemical measuring signals.However,the general wavelet-denoising methods do not effectually clear away the heteroscedastic noise in some chemical signals.In this contribution,the segment threshold wavelet denoising(STWD) method is proposed to eliminate the heteroscedastic noise that arises in the chemiluminescence detection signal of micro-fluidic electrophoresis chips.In the STWD,the wavelet denoising thresholds are independently chosen one segment by one segment according to the noise variances.Two synthetic signals(one with the general white noise,another with the heteroscedastic noise) were used to verify the validity of the STWD when the different parameters of wavelet denoising were adopted,and the STWD with optimized parameters was also applied to denoise the chemiluminescence detection signals of micro-fluidic electrophoresis chips.In comparison with the general wavelet denoising method,the STWD showed the better denoising effects in the two synthetic and the chemiluminescence signals.It indicates that the STWD could considerably improve the denoising performance of the signals with heteroscedastic noises.

Key words: Wavelet denoising, Segment threshold, Heteroscedastic noise, Chemiluminescence detection, Micro-fluidic electrophoresis chip