Observing the Impact of Critical Faults in Arithmetic Circuits for Digital Image Processing

Image processing applications are generally regarded as error-resilient, i.e., even an erroneous result is mostly considered sufficient. This is why many research works have focused on building low-power solutions which benefit from the inherent error-resilience of such applications. However, this should not be taken to mean that fault-resilience is unimportant in the context of image processing application scenarios. In this short demo, we highlight the impact of single faults in arithmetic circuits in select image processing applications. Arithmetic circuits are a class of digital circuits widely used in image processing. We show that even a single fault can be critical to application success as results demonstrate that a single fault can cause significant output error magnitude unacceptable for the application at hand.