In order to guarantee the precision of hydroxyl (OH) tagging velocimetry measurement in combustion flow field, we investigate a background noise removal method. Based on the analysis of noise characteristics of hydroxyl tagging velocimetry experimental data in combustion flow fields, we construct a numerical model of noise attenuation. For the background noise such as local combustion OH fluorescence interference and flow stray light, we adopt a spatial filtering method based on Hough transform. As for the physical, electrical, optical and sensor noises of the measurement system, we present a noise removal method of wavelet transform to improve signal-to-noise ratio. The background noise suppression method combining spatial filtering and wavelet transform is proposed to suppress system noise effect on the spatial filtering algorithm precision, and optimize the spatial filtering results. The research results show that the peak signal-to-noise ratio is improved by 16.79 dB, and signal-to-noise ratio is improved by 13.91 dB after imaging processing. The experimental data of combustion flow field are processed, which effectively suppresses background noise, achieves the effect of image preprocessing, and meets the requirement of measurement accuracy of laser diagnostic system for measurement accuracy.