Based on force sensing resistor(FSR) sensor, we designed insoles for pressure measurement, which were stable and reliable with a simple structure, and easy to wear and to do outdoor experiments with. So the insoles could be used for gait detection system. The hardware includes plantar pressure sensor array, signal conditioning unit and main circuit unit. The software has the function of data acquisition, signal processing, feature extraction and classification function. We collected 27 groups of gait data of a healthy person based on this system to analyze the data and study pressure distribution under various gait features, i.e. walking on the flat ground, uphill, downhill, up the stairs, and down the stairs. These five gait patterns for pattern recognition and classification by K-nearest neighbors (KNN) recognition algorithm reached up to 90% accuracy. This preliminarily verified the usefulness of the system.