| Summary: | 碩士 === 玄奘大學 === 資訊管理學系碩士班 === 105 === Detecting people’s position in images is challenging. From the past research, mostly histograms of oriented gradients (HOG) and support vector machine (SVM) have been applied for people detection. The approach consists of variables of geometric and photometric, whereby the detection result will not be affected as long as the people able to maintain the up-right posture or due to some minor posture changes. Fixed camera was used previously for people detection, and it was installed on vehicles for driving assistance. In recent years, it has been installed on unmanned aerial vehicle for better capturing view so that the images have more application mode. However, this method has some shortcomings whereby the detection result could be easily affected when there are some obstacles. Hence, we use thermal imaging camera to capture images then calculate heat source distributed location on thermal images and conduct visible light image comparison in order to find out people’s location. This proposed method will not affected by obstacles and able to detect people’s position accurately.
|
|---|
