Investigates Image data fusion techniques that combine image and track data from multiple sensors to achieve improved accuracies and more specific inferences than could be achieved by the use of a single sensor alone. Our aim is to explore the state-of-the-art image processing algorithms for achieving effective data fusion as in:
Object Detection on Low Resolution Environment
작성자 관리자날짜 2021-04-08 14:52:21조회수 26
- Object detection of a known class is a fundamental problem of computer vision. Many algorithms are proposed to solve this problem. Among those algorithms, appearance-based method is generally used because it has higher performance and faster calculating time than other methods. Low resolution images have the advantage of low memory consumption. But the quality of low resolution images are not enough for object detection. For the reason that current detecting method isn’t suitable for object detection on low resolution images, robust object detection algorithm for low resolution images is needed.
2. Main algorithm and principle
- For object detection on low resolution images, it is important that finding suitable feature which can describe character of object from limited information. Our method use EOH feature (Edge Orientation Histogram) for face detection and Haar-like feature for license plate detection. For increasing detecting speed, we use a cascade structure of classifier.
Fig.1) detecting diagram
3. Demo System - Object detection system is implemented by Visual C++ 2008 with OpenCV library.