[SCIE-MDPI Sensors] Deep Non-Line-of-Sight Imaging Using Echolocation - Seungwoo Jang , Ui-Hyeon Shin, Kwangsu Kim
- 인공지능 융합 연구실
- 조회수250
- 2022-11-14
Abstract. Non-line-of-sight (NLOS) imaging is aimed at visualizing hidden scenes from an observer’s(e.g., camera) viewpoint. Typically, hidden scenes are reconstructed using diffused signals thatemit light sources using optical equipment and are reflected multiple times. Optical systems arecommonly adopted in NLOS imaging because lasers can transport energy and focus light over longdistances without loss. In contrast, we propose NLOS imaging using acoustic equipment inspired byecholocation. Existing acoustic NLOS is a computational method motivated by seismic imaging thatanalyzes the geometry of underground structures. However, this physical method is susceptible tonoise and requires a clear signal, resulting in long data acquisition times. Therefore, we reduced thescan time by modifying the echoes to be collected simultaneously rather than sequentially. Then, wepropose end-to-end deep-learning models to overcome the challenges of echoes interfering with eachother. We designed three distinctive architectures: an encoder that extracts features by dividing multichannel echoes into groups and merging them hierarchically, a generator that constructs an imageof the hidden object, and a discriminator that compares the generated image with the ground-truthimage. The proposed model successfully reconstructed the outline of the hidden objects.
Keywords. non-line-of-sight; acoustic sensing; depth estimation; deep learning