Abstract
The planned deployment of turbines in the Florida Gulf stream could disturb the pattern of sea-turtle migration and reproduction. Therefore, a comprehensive study of the distribution of sea turtles is essential. Aerial counts and associated tasks are currently performed using human observations and distance sampling theory. The results of the observations made by human observers during flights are error-prone due to visibility biases. Video recordings have been introduced to circumvent some of the problems mentioned above and ensuring a systematic approach, frame-by-frame if needed, to minimizing count errors. The process is still time-consuming, tedious, and error prone, due to observer's fatigue. The goal of the proposed work is to improve such methodology by approaching the problem from an automated video analysis perspective. This work focuses on the first step in automating aerial surveys: design and development fo video acquisition system for aerial surveillance. The proposed solution consists of camera mounted on airplane which captures high resolution images at up to 15 frames per second and the captured images are to be compressed and saved to data recorder in real time. GPS is also integrated with the system to tag the images with GPS locations for biologists to crosscheck turtle sightings. This work also evaluates image compression schemes and the tradeoffs in compression performances based on computing required and storage capacity that will help in compressing the recorded data in real time.