Abstract: The ash content of flotation tailing is an key product indicator in flotation process. Therefore, the study and application of the technology for online detection of this indicator constitutes an important basis for realizing intelligent closed-loop control of flotation process. In order to address the problems encountered in detection of flotation tailing ash content with the γ-ray/X-ray and machine vision-based methods in current use, a novel diffusion reflection-based-detector together with the corresponding grey-level and multi-section linear interpolation-based ash prediction method are developed. Application shows that the detector uses a circular near-infrared light source with a wavelength of 850 mm as fill-in lighting for its good working effect; as the focus length and apertures of camera are both adjustable,clear images can be obtained, and only those images below the surface of the pulp are analyzed for eliminating the interference caused by foams flotating on pulp surface; the detected gray levels are in positive correlation with those of the pulp samples with different ash contents, and the grey level of the same pulp sample obtained through repeated measurements has a reliably high reproducibility; the linear interpolation function model constructed based on relation of sample ash with pulp image can make a relatively accurate prediction of the ash values of different slurry samples within the experimental range; and the prediction error of over 95% is ±1% with the maximum arithmetic average being 0.51%. The study made in the paper can help promote the realization of closed-circuit control of flotation process.