Improved Productivity of Mosaic Image by K-medoids and Feature Selection Mechanism on a Hadoop-Based Framework
Jau-Ji Shen, Chin‐Feng Lee, Kun-Liang Hou · 2016
In recent years, the development of sensor technologies and network technologies has rapidly improved, generating high volumes of data every second. To deal with the massive volumes of data created from the Internet of things (IoT) environment, the open source framework, Hadoop, has been developed and widely used in every domain. This study proposes a method improved the efficiency of K-Medoids algorithm by using the image feature selection mechanism as image metrics. The image feature selection mechanism also can find a better clustering result. Thus proposed method not only improved the efficiency of the codebook which output by K-Medoids algorithm but also improved the quality of mosaic image directly due to a better K-Medoids codebook.