“ Development of an airborne system for automatic colorbased object tracking

For many motives in the filmand picture producing industry helicopters are used to capture spectacular image sequences. Using helicopters implicates a high organization effort and high cost for manpower and material. Further on, helicopters are not easily available at remote places. Motivated by these problems the goal of this Diploma Thesis is to develop a model helicopter based system for automatic tracking and following of an object. This system simplifies helicopter based photo and film shootings, because it can be operated by a single person. Since model helicopters are effortlessly transportable, the system significantly reduces costs, organizational effort and solves the problem of availability. An automatic, image based tracking system, which is mounted on a model helicopter, is described in this work. The algorithm combines color histograms as similarity measures with a particle filter and achieves fast and robust tracking results. The system keeps the object of interest automatically in the viewing field of the camera so that the only task left to the camera operator is to initiate the imaging process. This simple task can be additionally managed by the helicopter pilot so that a single person is sufficient to steer the helicopter and to take the images. Apart from the model helicopter, the system can be divided into three parts: The object identification, the object tracking and the controlling of the camera which follows the object of interest. The target object is identified using a low quality digital camera mounted on a pan-tilt-unit (PTU). The image processing, which is necessary for the object identification and the controlling of the PTU is handled by an embedded PC. To produce hight quality film and photo material an additional professional camera is mounted on the PTU. The novel method of Integral Histogram images for fast histogram computation is proposed. This method enables the computation of many histograms in one image in constant time, independent of the size of the individual histograms. Because standard time discrete control algorithms depend on constant processing times, combining the Integral Histogram approach with a particle filter algorithm makes a successful controlling of the PTU possible.