Alzheimer's disease (AD) is the most common form of dementia. The symptoms include cognitive impairment and gait disturbances. Gait analysis reports showed that people with AD have slower gait speed, shorter step lengths, and lower step frequency. It was suggested that gait analyses may be important for therapies and diagnosis. However, recent gait analysis reports still involved floor sensors and wearable sensors, which may not be convenient for elderly patients. In this report, we proposed the use of Kinect™ device: a machine vision gait analysis approach. This device has been used for rehabilitation purposes in elderly care and stroke rehabilitation. Furthermore, it has been used in neurodegenerative disease studies. In this report, we would like to extract the gait components of a person based on the Kinect™ 25-joint system. It started with two subjects walking down a measurement aisle 5 times. The recorded videos of the 2 subjects were subsequently interpreted into the position of joints over time following the Cartesian coordinate. Next, we extracted the gait components manually from the joints involved in determining gait components (left foot & right foot) using a Microsoft Excel software. The result validation was done using t-tests comparing the gait components of the 2 subjects. It was shown that the results were valid, which opens an opportunity to use this device for AD therapies.