Today's, blanched or roasted hazelnuts are widely used in chocolate and biscuits industry. Partly skin removed hazelnuts kernels are caused to deterioration in flavor and taste of the chocolate. Many of chocolate and biscuits cooperation in Turkey are using traditional manual selection methods. However, manual selection has high error rate, time consuming and costly. In this study, the partly skin removed hazelnut kernel, skin removed and rotten hazelnuts kernels which are obtained from processed kernels are classified by using computer vision approach. In the proposed study, the processed hazelnut kernels are classified with 93.57 % classification accuracy rate.