This research aimed at providing a quick, and non-destructive method for estimating freshness of intact Japanese dace (Tribolodon hakonensis) fish refrigerated below 5 degrees C using multispectral imaging technique coupled with multivariate analysis techniques. The fluorescence excitation-emission matrices (EEMs) for outside parts of the fish (eyeball, scales) were taken to establish the proper excitation wavelengths. Afterwards, four lighting devices; white, 395 nm, 365 nm and 280 nm LEDs were used to illuminate the sample for capturing images using both the UV camera and a common color camera. Biochemical analysis (electrophoresis measurement) was carried out to examine fish freshness and then expressed using a reference method as K values. Both EEM and imaging data were modelled with the measured K value using partial least square (PLS) regression. A novel algorithm based on multispectral imaging data was proposed as a potential fish freshness indicator during storage. From the results, R2 of 0.94 and RMSE of 2.42% were achieved. This research demonstrated that the fluorescence multispectral imaging technique is a powerful tool with great potential in non-destructive monitoring and examining fish freshness during storage.