Abstract: Based on Retinex theory,an evaluation model for measuring gloss of people's facial skin was established.Via the simulation experiments on samples of 196 volunteers,a library of people's facial skin images was built,which was accessed by means of single scale Retinex algorithm to transform the readings of people facial skin images S(x,y) into reflectance images R(x,y).By this way,the mean and variance of pixels of R(x,y),which were processed in normalizing with the gloss reading given by gloss meter and the error analysis was conducted by use of relative square error method.The experimental results indicated that the evaluation on facial skin gloss by R(x,y) pixel mean method agrees with that measured by the gloss meter with an error ratio of less than 0.13.Therefore,the method can be deemed as a suitable,simple and effective way to evaluate people's facial skin gloss.

Key words: facial skin, Retinex algorithm, gloss