HandSee

DeepFisheye

“…There fore, most near-surface finger tracking technologies for mobile devices used a camera to track single or multiple fingers. Re cently, Yu et al [75] suggested HandSee, that tracks fingertips' 3D locations using a smartphone with a prism attached to its front camera. The prism is used to acquire a reflected image from the touchscreen.…”

“…index, middle) from a hand image difficult. HandSee's [75] approach failed to cover the whole touchscreen area owing to the camera's limited FOV. Furthermore, owing to the prism's structure, the sensing volume near the prism was very narrow.…”

“…The nearsurface multi-finger tracking (NMFT) technology tracks the 3D location of each fingertip near an interactive surface. Al though it enables mid-air [13,61,75] and finger-aware inter actions [61,75,79], there is no practical solution for NMFT that can be applied to mobile touchscreen devices. An NMFT system should satisfy the following requirements.…”

“…Visionbased approaches are popular for NMFT because they are more user-friendly than the wearable systems such as mag netic field sensing [9,11], wherein users are required to wear devices on their hands. However, the hardware configurations of vision-based approaches [20,61,75] are not compact, be cause the camera, which has a limited field of view (FOV), must be distanced from a the hand to capture its sufficient portion. Yu et al [75] attached a prism to the front camera of a smartphone to solve this problem; however, their solution could not distinguish and track multiple fingers.…”

“…However, the hardware configurations of vision-based approaches [20,61,75] are not compact, be cause the camera, which has a limited field of view (FOV), must be distanced from a the hand to capture its sufficient portion. Yu et al [75] attached a prism to the front camera of a smartphone to solve this problem; however, their solution could not distinguish and track multiple fingers.…”