An Integrated Lock-in Amplifier and Near Infrared Method for Finding Fruits Fully Behind Leaves in a Homemade Testbed

Detecting fully occluded objects is of interest for various practical problems, such as harvesting and yield prediction in farming, which are physically demanding and heavily labor-dependent.Many approaches have been explored by researchers aiming to solve this problem.However, they are ineffective due to inherent challenges: the strength of signals reflected from hidden objects is weak, and those signals are always buried in high-magnitude noise.In this study, a method combining nearinfrared (NIR) and lock-in-amplifier (LIA) techniques is proposed to tackle these challenges.Two questions are answered.Can a fully covered fruit be detected purely based on reflected NIR signals?Can LIA extract reflected signals from high-magnitude noise?This study addresses these questions from theoretical and experimental points of view, including NIR photon particle propagation, LIA in the image format, low-cost experiment apparatus, etc.In total, 268 videos were collected over 134 valid experiments with tomatoes and cucumbers as objects.Both alternate hypotheses were validated and answered.