In this study, Palm Glove, an interface device positioned on the palm, is developed as a wearable device that can recognize finger motion. An algorithm to recognize thumb-to-finger tap gestures for the device is also proposed. Palm Glove can obtain stable finger images even in case of wrist movement, as the camera in the device is positioned on the palm. The thumb-to-finger tap recognition algorithm uses projected extreme point (TTRPP), which is a robust and light algorithm that detects thumb-to-finger tap gestures by projecting two-dimensional geometric image features of fingers onto one-dimensional points. A database was constructed using the proposed hardware and algorithm and a 99.22% recognition rate was achieved. Furthermore, the proposed hardware was evaluated through usability assessment to be used as a practical real-time interface device.