Bioinspired figure-ground discrimination via visual motion smoothing

Flies detect and track moving targets among visual clutter, and this process mainly relies on visual motion. Visual motion is analyzed or computed with the pathway from the retina to T4/T5 cells. The computation of local directional motion was formulated as an elementary movement detector (EMD) model more than half a century ago. Solving target detection or figure-ground discrimination problems can be equivalent to extracting boundaries between a target and the background based on the motion discontinuities in the output of a retinotopic array of EMDs. Individual EMDs cannot measure true velocities, however, due to their sensitivity to pattern properties such as luminance contrast and spatial frequency content. It remains unclear how local directional motion signals are further integrated to enable figure-ground discrimination. Here, we present a computational model inspired by fly motion vision. Simulations suggest that the heavily fluctuating output of an EMD array is naturally surmounted by a lobula network, which is hypothesized to be downstream of the local motion detectors and have parallel pathways with distinct directional selectivity. The lobula network carries out a spatiotemporal smoothing operation for visual motion, especially across time, enabling the segmentation of moving figures from the background. The model qualitatively reproduces experimental observations in the visually evoked response characteristics of one type of lobula columnar (LC) cell. The model is further shown to be robust to natural scene variability. Our results suggest that the lobula is involved in local motion-based target detection. Author summaryFlies are adept at target tracking during high-speed aerial maneuvers, although their brains are tiny. Along the pathway from the retina to the deep optic lobe, local directional motion at different positions within the visual field is first detected, based on which optic flow the fly experiences is then encoded. In contrast to extensive research on optic flow processing in the fly visual system, much less is known regarding how target detection is performed, even under the simple condition of a tethered fly. In this study, we built a fly inspired model to investigate how a textured figure is discriminated from a similarly textured background by relying on relative motion cues alone. We found that the secret involved in a network downstream of local motion detectors extracting the figure is to naturally exploit the temporal coherence of the moving figure. Without a response integration in time to a sequence of temporally ordered input frames, the downstream network fails to extract the figure, even if it has made full use of spatial coherence between adjacent pixels in individual frames. The modeling is helpful for understanding where and how target detection is performed in the visual system of flies. The result could also be inspiring for motion detection in machine vision systems that rely only on a single camera.