Determination of salmon freshness by computer vision based on eye color

To evaluate and predict the freshness of salmon nondestructively, a computer vision technology was developed based on eye color to predict multiple freshness indicators of salmon simultaneously during storage at 0 degrees C. The RGB, L*a*b* , and HSI color spaces of eye images were analyzed by an image processing algorithm. It is demonstrated that the eye color parameters R, G, B, L* , I, and Delta E were correlated with freshness indicators to establish the multiple linear regression (MLR) and support vector regression (SVR) models. The MLR models outperformed SVR models with high correlation coefficients R2, F value, and low relative errors. The achieved results showed that it was a nondestructive, fast method for predicting the freshness of salmon stored at 0 degrees C by evaluating the eye color parameters with computer vision.