| Summary: | 碩士 === 國立中興大學 === 資訊管理學系所 === 104 === Digital images are non-structure data. In order to apply some image processing techniques on a digital image, we need high computing power. Nowadays, digital images have become an issue of Big Data. In our research, we decide to implement a feature-based K-Medoids on a popular Big Data analysis tool, Hadoop. Hadoop can provide high computing power. The algorithm of mosaic image is a high computing complexity process. Especially in Big Data of image environment, the method needs to deal with about tons of images. There are three main goals of our research. Frist, we want to use an unsupervised clustering method, K-Medoids, to cluster the image dataset and build a codebook. Then we can use the codebook to generate the mosaic image to reduce the processing time. Second, we use two feature selection metrics to develop a derived K-Medoids methods, called feature-based K-Medoids. Feature-based K-Medoids can cluster the image dataset faster by the feature selection mechanism. Third, our method surely reduces the processing time of mosaic image by the codebook. Though the image quality of our method is slightly lower than Szul et al.’s method, our method retains an acceptable image quality.
|
|---|
