Image Retrieval with Fusion Descriptor from Deep Learning and Compressed Domain Features

• Main Authors: Che-Yi Wu, 吳哲逸
• Other Authors: Jing-Ming Guo
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/6f6992

碩士 === 國立臺灣科技大學 === 電機工程系 === 104 === This thesis presents an effective image retrieval by combining low-level features from Dot-Diffused Block Truncation Coding (DDBTC) and high-level features from Convolutional Neural Network (CNN) model. The low-level features are constructed by the proposed two-layer codebook feature from DDBTC bitmap, maximum, and minimum quantizers. The two-layer codebook is to improve the limited dimension of original codebook. The high-level feature is from CNN which is a very effective approach for deep learning. The high-level feature has been widely applied in recognition and classification, and it is also regarded as close to human perception. With the fusion of the DDBTC and CNN features, the extended deep learning two-layer codebook features (DL-TLCF) is generated using the proposed two-layer codebook, dimension reduction, and similarity normalization to improve the overall retrieval rate. Two metrics, average precision rate (APR) and average recall rate (ARR), are employed to examine various datasets. As documented in the experimental results, the proposed schemes can achieve superior performance compared to the state-of-the-art methods with either low- or high-level features in terms of the retrieval rate. Thus, it can be a strong candidate for various practical image retrieval related applications.