3rd TAGRA

In modern video surveillance systems, the large area under observation and the varying distances of objects from the camera may reduce the informational value of the captured images. In particular, small and distant objects are represented by a limited number of pixels within the camera frame, making their detection and subsequent analysis more difficult. To address this problem, instead of relying solely on a camera operating in a fixed position, it is appropriate to enable the camera to automatically change its orientation according to the movement of the observed object and, when necessary, perform optical zooming. This study presents a system approach based on the detection of moving objects from image sequences, determination of their direction of movement, automatic orientation of the camera toward the detected object, and automatic optical zooming depending on the object's size within the frame. Following the initial image preprocessing stage, differences between consecutive frames are calculated to identify moving regions, while morphological processing is applied to reduce the influence of incidental image changes. For each detected object, a characteristic tracking point is determined, and the displacement of this point relative to previous frames is used to control the camera's movement. When the object's movement and its size within the image satisfy predefined conditions, optical zooming is activated. Such an approach enables distant and small objects to be maintained at a more suitable size within the image, thereby facilitating their observation and subsequent analysis.