For the demand of inland water quality monitoring, a ground-based multi-spectral imaging method has been developed. By means of developing an instrument which can gather the multi-spectral data of the waterbody, the method can be used for real-time monitoring the contamination of inland waters, such as cyanobacteria bloom and phytoplankton. The research is focused on the technology of high-resolution multi-spectral data extraction and the theory of contaminant inversion model. Four branches of light beams of simple spectrum are obtained with spectral filters and are recorded by four groups of lens and detectors respectively. The four interested wavelength is chosen as 565 nm, 620 nm, 660 nm, 750 nm, according to the typical reflection peaks and dips of the contamination with a spectral resolution of 15 nm. The optical design features a field of view of 25.2×19.3 degree with a 16mm focal lens. The camera's resolution is 1628×1236 with the pixel size of 4.4 microns that reaches the spatial resolution of 0.945 arc min. The multi-spectral image is obtained through out-door experiments by monitoring the inland lake-Dianchi at a distance of 5 kilometers. After data revision, we can identify the constituent of the underwater contaminant and explain the pollution situation of cyanobacteria bloom in a certain period quantitatively. The inversion and extraction accuracy can reach at least 85%. And the long-term observation can explore the seasonal pattern of cyanobacteria bloom outbreak. © 2019 SPIE.