Hand functions such as hand opening, hand closing, and object grasping are needed to perform many tasks in daily life, but developing effective rehabilitation devices for hand function has been a challenge. To encourage stroke survivors to use residual motor abilities in the rehabilitation process, we developed a hand rehabilitation robot control interface using surface electromyograph in stroke patients, and compared whether patients' actual performance of motion improves when utilizing this interface. We obtain 72% level of classifier accuracy, with an average task operation time of 4.5 seconds decreasing. It is expected that the proposed control interface will provide a high level of exercise recovery in the future.