Optical camera communciation (OCC) system is a wirless communication system that uses camera as receiver. In the non-line-of-sight(NLOS) OCC system, smartphont camera can capture the image background and light reflected by the object. The NLOS OCC system using reflected light exhibits its advantages, such as (i) reduced requirement on alignment and (ii) the expanded received rolling shutter pattern to achieve a larger data rate. However, there are few investigations and solutions to the effect of unconcise handshaking in the NLOS OCC system. In this project, we proposed an anti-shaking algorithm to improve communication performance for NLOS OCC system.
In the NLOS OCC system, we first capture an image with long exposure time. After that, we decrease the exposure time and capture a video to record light signals. Due to rolling shutter effect, light signals are recorded as black and white strpes overlapped on the image background. To alleviate the interference of image background, we use long-expousre image for signal scaling. However, the handshaking generated by the mobile phone user will cause misalignment between the long exposure image. After scaling, the residual amplitude fluctuation will severely degrade the BER performance. Therefore, the challenge of anti-shaking algorithm is to realize image registration between the long-exposure image and the corrupted short-exposure images.
Match the key points based on their features. The features may be corrupted by the stripes.
Estimate relative translative offset between two similar images. Exhibit robust performance for noisy and corrupted images. The perspective transformation cannot be derived.
Step 1: Find key points from the long-exposure image. Step 2: i) Apply phase correlation algorithm to the image blocks with key points at the center. ii) Calculate perspective transformation matrix based on the positions of key point pairs. iii) Generate transformed long-exposure image.
Anaconda 3
Matlab 2021b
0.jpeg: Image captured with long exposure time
1618053369425.mp4：Video captured with low expsoure time
z2_frame_saved.m: Convert video to images
anti_shaking.py: Anti-shaking algorithm
mutual_information.py: Calculate mutual information between the low=exposure image and long-exposure image
Liqiong Liu, Department of Information Engineering, The Chinese University of Hong Kong
Distributed under the MIT License.
[1] L. Liu and L. -K. Chen, "Li-poster: Real-time Non-line-of-sight Optical Camera Communication for Hand-held Smartphone Applications," in 2021 Optical Fiber Communications Conference and Exhibition (OFC), 2021, pp. 1-3.