Image reconstruction using single-pixel camera via compressed sensing
Mr. HAO Yifeng
PhD candidate in the Mechanical Engineering Dept.
Date & Time
Tuesday, 3 May 2022
Typical cameras based on CCD or CMOS digital technologies use an array of millions of detectors to capture the image. In contrast, the single-pixel camera is able to reconstruct the images using a single-pixel detector, providing high resolution and cheap price for imaging at non-visible wavelengths. This talk will cover the principle of the single-pixel camera, including both the hardware setup in aspects of modulation schemes, mask design, and the reconstruction algorithms. A sequence of mask patterns, generated by modulation technologies like the digital micromirror device (DMD), filter the target scene, and the overall intensity is captured by the single-pixel detector. The image of the scene can be reconstructed using fewer measurements than the overall pixels via compressed sensing theory, in which way we can achieve sub-Nyquist image acquisition. With its unique characteristics, the single-pixel camera holds potential for a variety of applications, such as 3-D image reconstruction in non-visible wavelengths and special environments. The recent applications and the latest research outcome will also be discussed in this talk.
Robotics and Control