| Summary: | 碩士 === 逢甲大學 === 資訊工程學系 === 104 === In recent years, the technology and application of the Internet of Things (IoT) become more and more popular. Everything can connect with each other on the network. Especially for IP cameras, automatic image analysis, object recognition and objects tracking become more and more commonplace. With the advance of science and technology, resolution of the captured image is increasingly high. Therefore, how to real-time process images has become an important issue.
In the thesis, we design two efficient intelligent identification libraries to process image recognition and objects tracking. We take two frequently used algorithms, Background Subtraction and Template Matching, as the basic approaches in the library. Then, we integrate some complex but frequently used features with these two approaches. In addition, we also enhance the performance of these two libraries by parallel computing technologies including OpenCL, GPU, multi-threading programming. Users can develop image recognition and objects tracking applications more easily and efficiently using the proposed modules.
|
|---|
