This paper presents a sensing and compression system for image detection, based on large-area electronics (LAE). LAE allows us to create expansive, yet highly-dense arrays of sensors, enabling integration of millions of pixels. However, the thin-film transistors (TFTs) available in LAE have low performance and high variability, requiring the sensor data to be fed to CMOS ICs for processing. This results in a large number of interconnections, which raises system cost, and limits system scalability and robustness. To overcome this, the presented system employs random projection, a method from statistical signal processing, to compress the pixel data from a large array of image sensors in the LAE domain using TFTs. Random projection preserves the information required for subsequent classification, and, as we show, is highly tolerant to device-level variabilities and amenable to parallelized implementation. The system integrates an amorphous-silicon (a-Si) TFT compression circuit with an array of a-Si photoconductors, representing an 80 × 80 active matrix, performing up to 80× compression of the 80 signal interfaces. For demonstration, image classification of handwritten digits from the MNIST database is performed, achieving average error rates of 2-25% for 8-80× compression (e.g., 7% at 20× compression).
|Original language||English (US)|
|Number of pages||12|
|Journal||IEEE Transactions on Circuits and Systems I: Regular Papers|
|State||Published - Nov 1 2016|
All Science Journal Classification (ASJC) codes
- Electrical and Electronic Engineering