The aim of the presented work was the development of software technique for detection and tracking of moving objects in video sequences. It is intended to serve as an automatic video surveillance or traffic control system. Local image features are detected and tracked in the presented system. Two clustering algorithms are utilised for this task succes-fully. Firstly, the QT (Quality Threshold) algorithm has a potential of new object detection. Secondly, modification of a well known K-means algorithm proved its usefulness in tracking moving objects in image sequences. For reduction of the analysed data, corners are detected in consecutive images. Their motion vector and coordinates produce feature vectors for an image classifier. The obtained results show the ability of the proposed technique to detect and track multiple objects on the basis of their local, visual features. No model matching technique was necessary, which simplified overall approach. Comparatively low number of operations, required to perform tracking process, gives the possibility to implement the algorithm in real time on modern graphics processing unit in PC computers.
2
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
To transform images with a caption into images without one or to replace the current captions by a different one, we sometimes need to recover original image for caption areas. Manual processing is possible when the amount of recovery to be done is small. However, as the amount of video to be processed increases , manual processing is impossible or takes too much time. Conventional researches on image restoration have focused on restoring blurred images to sharp images using frequency filtering or video coding for transferring images. This paper proposes a method for recovering original images using camera motion and video information such as caption regions and scene changes. The method decides the direction of recovery using the caption information (the start and end frames of caption) and scene change information. According to direction of recovery, a rough estimate of the direction and position of the original image is obtained using calculated motion vector from camera motion. Because the camera motion dose not reflects local object motion, some distortion can happen in the recovered image. To solve this problem, BMA (Block Matching Algorithm) that is applied in units of caption character components on the obtained recovery ppositions. Experimental results show that the case of images having little motions is well recovered. We see that the case of images having motion in complex background is also recovered.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.