Here's a crazy idea:
Monitor the scattered light from two lasers. Use the intensity of light, as read by the two photo sensors, to determine the position of the finger. Each sensor is behind a tube to limit ambient light influence. The tube's diameter and depth would be tailored to favor laser scattered light over ambient light. Perhaps a third photo sensor could monitor ambient light and contribute corrections to the calculation.
With experimentation one could arrive at a "Scattered Light Intensity/Finger Position" profile. Perhaps this could lead to a mathematical solution OR perhaps some sort of look-up table -- perhaps along with extrapolation?
I would make the inside of the tubes as black as possible, to reduce reflections. That way, it can be designed to "see" only direct scatter from the finger.
Also, to deal with the variability of ambient light, the lasers could be modulated and the photo sensors would, when the laser is off, then be used to measure the ambient light, and this could be subtracted.
Variations in "finger reflectivity" could be an issue. But, perhaps the information from the two lasers could be used to balance this out. For instance, the profile could only involve the ratio of intensity from the two sensors. OR, the musician could be provided with a special glove, to unify the reflectivity. But, finger size might, also, be a factor -- but not as much, and looking at ratios should remove that variability.