Detection of mental task in Brain-Computer Interface (BCI) is a crucial issue for different application, especially rehabilitation. As known, frontal lobe the largest cortex of brain is involved in function like thinking, planning and organization. The aim of study is detection of motor imagery-based hand grasping. Thus, channels located on frontal lobe are focused on. Laplacian reference channel is obtained from electrodes located on frontal lobe to decrease number of channel and complexity. Classification of motor imagery-based hand grasping highly is up to feature extraction from signal. In this study, new features related to cross-correlation coefficients of EEG waves (delta, theta, alpha, low beta, high beta) in frequency domain are proposed. They are utilized as effective features. It is observed that mean accuracy of motor imagery-based right hand grasping classification is 71.50%.