基于android平台的二进制特征方法的研究-软件工程专业论文.docxVIP

ABSTRACT With the rapidly growing popularity of mobile devices and increasing image data sizes, binary features has become a new hotspot since the traditional feature methods use high dimensional features of floating-point type, and cost a lot of storage space and computing time. They are able to quickly compute and match with efficient storage. BRIEF(Binary Robust Independent Elementary Features), ORB (Oriented FAST and Rotated BRIEF), BRISK(Binary Robust Invariant Scalable Keypoints) and FREAK (Fast Retina Keypoint) these four binary features were researched in this paper while SIFT(Scale Invariant Feature Transform) and SURF(Speeded Up Robust Features) as references. Firstly, the general characteristics of binary features were researched in feature detecting, description and matching. The characteristics include sampling models and robust. The space and time performance was studied both on mobile device and PC. Also, the robustness of binary features under scale change, rotation, blur, illumination change and perspective change was researched. The different stages of feature methods and their performance bottleneck were analyzed. The performance and platform influences and limitations were studied based on the mobile platform. Finally, binary feature methods were paralleled to improve the performance. Experiments and analysis were done on the automated evaluation system, which was designed and implemented on both PC and Android platforms. Evaluation system was extended with new metrics. The test image dataset was expended. The results show that binary features have superior performance in time and space that the efficiencies are improved about 10 times. Also they have very good robustness overall which is comparable with SIFT’s. Parallel can achieve a speedup of 1.3 or more, which can improve the performance. KEY WORDS：binary features, BRIEF, ORB, BRISK, FREAK, performance study 目 录 HYPERLINK \l _bookmark0 第一章 绪论 1 HYPERLINK \l _bookmark1 1.1 研究背景及意义 1 HYP