Recording dynamic facial micro-expressions with a multi-focus camera array.

Abstract

We present a multi-camera array for capturing dynamic high-resolution videos of the human face. Compared to traditional single-camera configurations, our array of 54 individual cameras allows stitching of high-resolution composite video frames (709 megapixels total). In our novel multi-focus strategy, each camera in the array focuses on a unique object plane to resolve non-planar surfaces at a higher resolution than a standard single-lens camera design. By overcoming the standard resolution and depth-of-field (DOF) tradeoffs, we use our array design to capture video of macroscopically curved surfaces such as the human face at a lateral resolution of 26.14 ± 5.8 µm across a composite DOF of ∼43 mm that covers the entire face (85 cm²+ FOV). Compared to a single-focus configuration, this is almost a 10-fold increase in effective DOF. We demonstrate how our multi-focus camera array can capture dynamic facial expressions at microscopic resolution with relevance in several biomedical applications.