Abstract: In order to quickly and accurately non-destructive real-time monitoring of fish freshness, a visual colorimetric array sensor was constructed in this study based on anodized aluminum flakes as the substrate and 13 indicators selected as gas-sensitive materials. The freshness of turbot during storage was measured by using the constructed colorimetric array sensor. The results showed that the total volatile basic nitrogen (TVB-N) value of turbot reached 32.85 mg/100 g when the storage time was more than 22 h at 25 ℃, and the fish was rotten. At this time, the color of the array points in the colorimetric sensor array changed. The RGB values of sensor array points before and after response were extracted and subtracted, and the clustering analysis and principal component analysis were carried out. The results showed that the colorimetric array sensor was able to accurately distinguish the freshness of fish in different storage time periods. In addition, the array sensor was applied to detect the freshness of carp and American redfish at different storage times, and the results showed that there was no significant difference between the TVB-N values predicted by the colorimetric array sensor and the actual measured values. Therefore, it is feasible to apply the sensor to fish freshness monitoring.