CS 526 F20****15 Computer Graphics 2 : Weekly Research and Project Journal

---

I. Articles Summary

1. Davis, Abe, et al. "The visual microphone: Passive recovery of sound from video." ACM Trans. Graph 33.4 (2014): 79.
Article link
In this paper, the authors present a method to reconstruct sound from video footage recorded with high-speed camera, using computer vision on vibrating object in the footage. They compared different vibrating objects, and the object gives most accurate result is potato chip bag, since they vibrate most.

The authors evaluate the outcome result based on the intelligibilty, SNR (signal to noise ratio), and compare with the original version. They also explored the possibility of using a normal camera to achieve the same result. The sound recovery composes of 3 steps: 1. Compute Local Motion Signal, 2. Comput Global Motion Signal and 3. Denoising

---

2. Google Area Learning. Project Tango
Article link

A Google project used for mapping the environment. In this project, Google creates a device that can remember the feature of the environment. Of course, the location of the device is stored in itself, and when it moves back to the previous position, it remembers that it was there before. Useful for gaming, and indoor navigation, virtual reality in museum.
Use computer vision to make the device be able to understand their environment.
SLAM : Simultaneous Localization And Mapping.