| Summary: | 碩士 === 亞洲大學 === 資訊工程學系碩士班 === 101 === Image descriptors have been widely used in many computer vision applications. However, the image descriptors are sensitive to different imaging conditions such as changes in lighting geometry and illumination colors. If the image descriptors used for matching are no robust enough to cope with such changes, the recognition accuracy will be greatly affected. Therefore, the purpose of this study is to develop an efficient and robust image descriptor for real time and large scale image matching applications, and to evaluate its performance under different imaging conditions. Currently, the most widely used image descriptor is SIFT. However, SIFT descriptor demands large amount of calculation thus it is not suitable for real time processing and for low performance devices. The proposed descriptor uses binary operations to significantly reduce the cost for image gradient calculation that is required in SIFT, therefore increase the processing speed. We evaluated the performance of the proposed descriptor for matching images under different lighting and noise conditions. Experimental results indicated that the performance of the proposed descriptor was comparable to SIFT descriptor, whereas the computational complexity of the proposed descriptor is much smaller than SIFT.
|
|---|
