In recent years, China’s economy has developed rapidly, industrialization has become higher and higher, and atmospheric pollution has intensified, seriously affecting people’s daily lives. Therefore, real-time monitoring and research on atmospheric pollutants is particularly important. The interaction of various pollution sources in the atmosphere of the urban boundary layer makes the pollution problem complex and variable, especially the vertical distribution and change of pollutants in the atmosphere during heavy pollution. Imaging differential absorption spectroscopy (I-DOAS) is used to detect the spatial distribution of pollutants. The research at home and abroad is based on ground-based scanning, airborne and space-borne platforms. Because of its long-distance, multi-component, high-resolution and continuous real-time observation, the observation range can be extended from small scale to large area, which can provide important data support for analyzing the current situation of the atmospheric environment. Ground-based imaging differential absorption spectroscopy is generally used to detect a certain pollution source. This paper mainly studies its detection method for urban atmospheric boundary layer pollutant distribution. It introduces the principle of differential absorption spectroscopy (DOAS) based on Beer-Lambert law, and introduces the imaging principle of imaging system based on “push-broom”. Taking the common pollutant NO2 in the atmosphere as an example, on June 12, 2018, the imaging telemetry experiment of NO2 in the boundary layer was carried out in Science Island of Hefei City. The front end of the multi-core fiber bundle was coupled with the ultraviolet lens, and the back end was connected to the slit of the spectrometer. The ultraviolet lens was mounted on the two-dimensional rotating motor. Set the appropriate elevation angle of the two-dimensional rotating electric machine, and rotated it from 0° to 90° in the horizontal direction. The observation area included the suburb, power plant area and urban area. The zenith solar spectrum was selected as the reference spectrum, and the corresponding multi-channel spectra were combined and extracted for the measured spectrum and the reference spectrum. 38 spectra were obtained for each acquisition. Data inversion of all measured spectra was performed using the DOAS inversion method to obtain the differential slant column density (DSCD) of 38×90 NO2, and the density information was matched with the pixels on the spatial dimension according to the geometric model of the observation angle. After deducting the complex background, the two-dimensional distribution images of the NO2 differential slant column density in the boundary layer of Hefei City were obtained, according to the scanning direction. Compared with the MAX-DOAS data observed at the same time, the correlation coefficients of the two in the suburbs, power plant area and urban area were 0.86, 0.87 and 0.83, respectively. The results showed that the system can effectively obtain the distribution information of atmospheric pollutant concentration in urban boundary layer.