| Summary: | 碩士 === 國立宜蘭大學 === 電機工程學系碩士班 === 104 === In recent years, pedestrian detection and face recognition technology have developed rapidly. However, pedestrian detection portion still has much room for development.
Through the captured image, pedestrian detection can be applied to the vehicle drive assist system or monitoring system to recognize pedestrians, but the accuracy rate and speed of execution are also very important parts, so in this study, an iOS-based pedestrian recognition system has been proposed, the function of the pedestrian detection applied to mobile devices. This study will explore accuracy and speed of execution of pedestrian recognition supported by OpenCV HOG method on the iOS system and find ways to improve its performance. Using HOG method to turn images into gray scale, histogram equalization (Histogram Normalized), define the calculation region, the gradient calculation, after obtaining the feature vector classification with SVM classifier to do, and then compare with INRIA database which has trained, further determine whether there are pedestrians or not and mark the location. Then, add the Region of Interest to reduce the amount of computation, so we can achieve the effect of promoting implementation rate. Finally, we will analysis the results of the execution speed and pedestrian detection accuracy.
On the other hand, weather data application on mobile devices is also a popular topic that rapidly growing in recent years, furthermore, many meteorological information app has also been added to the App store. This study presents a graphical interface for weather information, so that we can get the weather information provided by the Central Weather Bureau through the URL, and these weather information presented to the user for viewing. In addition, the system will also store the weather information for produce to become a weather information database. Use Mac OS in the Core Data to store the data in an XML format file, so that the application can read past historical data at any time, and the future can also use them to do the simulation and analysis of historical data, and then do estimate the rainfall.
|
|---|
