Signalling the future

2 July 2019

Signal and image processing engineer Dr Stanley Chan (Croucher Scholarship 2008, Croucher Fellowship 2011) is powering forward with computational imaging methods for very low light situations, which has exciting potential to boost diagnostic capabilities in fields ranging from medicine to autonomous vehicles.

“Computational imaging is a very different concept from traditional image processing,” the Purdue University assistant professor explained. “It is the co-design of sensors and algorithms.” This opens the way to gather images that would be unobtainable with a conventional sensor or algorithm alone, he explained.

Such technologies have already yielded exciting results. In April 2019, the first images of a black hole made international headlines, with the daughter of Chan’s Purdue colleague helping to create the historic pictures by combining data gathered from eight Event Horizon telescopes.

However, building understanding of dead stars is just one of many applications for computational imaging. Autonomous vehicles, medical devices, and robotics are all within the sights of computational imaging.

While these areas seem highly diverse, they all share a common feature: the use of cameras. And it is cameras that form the focus of Chan and his research collaborators, who have proposed a third-generation digital image sensor to replace the charge-coupled devices (CCDs) and complementary metal-oxide semiconductors (CMOS) now on the market.

Sensors are characterised by size and number of pixels, camera speed, and the brightest and darkest spots they can capture, Chan said. Noise – random variations in brightness or colour data processed by the camera – is another important factor. Usually, sensors excel in one of these aspects but find it difficult to perform well in all of them.

We have a much bigger goal than trying to sell you a cell phone

Chan is seeking to make this possible. “We are designing a type of camera, empowered by special image processing algorithms, for very low-light situations where a typical iPhone [wouldn’t work],” he said. If he and his colleagues succeed, the new technology could ultimately make cameras of today obsolete.

However, Chan’s motivations are not commercial. “We have a much bigger goal than trying to sell you a cell phone,” he said. Instead he wants to construct images using only a few photons to allow scientists to carry out medical diagnoses and autonomous cars to detect objects within the darkest environments, for example.

By combining physics, mathematics, and computer programming, he expects this new sensor to be developed within 10 years. It’s an ambitious task, but one that fits perfectly with Chan’s objectives in life.

The young academic knew he wanted to be involved in scientific discovery and innovation while still at school. “I didn’t really know what I wanted to do exactly, but I just wanted to go to university and do science,” he said.

After completing a bachelor degree in electrical engineering at the University of Hong Kong, Chan moved to the United States to pursue postgraduate studies. He later secured a postdoctoral research fellowship at Harvard University before joining the faculty at Purdue University.

In the future, Chan sees machine learning as a key area to explore to link his work even closer to robotics: “We are sitting at the intersection of sensors and algorithms.” And with sensors playing a key role in improving the performance of robots, they will be instrumental as the field develops, he noted.



Dr Stanley H. Chan is currently an assistant professor at the School of Electrical and Computer Engineering and Department of Statistics at Purdue University, US. He received his PhD in Electrical Engineering and MA in Mathematics from the University of California, San Diego in 2011 and 2009 respectively, and BEng in Electrical Engineering (with first class honours) from the University of Hong Kong in 2007. Prior to joining Purdue, he was a postdoctoral research fellow at Harvard John A Paulson School of Engineering and Applied Sciences from 2012 to 2014. His research interests include signal and image processing, applied statistics, and large-scale numerical optimisation. He received a Croucher Fellowship in 2011 and a Croucher Scholarship in 2008.


To view Dr Chan’s Croucher profile, please click here.