SUMMARY
In this research, we will focus on facial expressions to detect customer satisfaction in mini markets where the service level is less than optimal. To find out the level of custome satisfaction can be seen through facial recognition tahen through CCTV in the mini market. The problems that occur are many customers who do not directly convey the impression that is felt when shopping, while minimarkets and shopping conters must know the level of customer satisfaction to improve sales strategies. Research to solve the problem is still rerely done, therefore one of the roles of intelligent computing is to solve the problem using Support Vector machine (SVM). The purpose of this study is to improve the accuracy of facial expressions of mini market customers through improved preprocessing. The results of the application of the otsu method and the gaussian function can be used for the preprocessing stage through a threshold image that has good image quality. The otsu-gaussian method is not effectively used for preprocessing data sourced from video or images with poor image quality, making it difficult to recognize faces.