Content-based Image Retrieval Using Effective Synthesized Images from Different Camera Views via pixelNeRF
Yuki Era, Ren Togo, Keisuke Maeda, Takahiro Ogawa, Miki Haseyama · 2022 IEEE 11th Global Conference on Consumer Electronics (GCCE) · 2022
We propose a novel image retrieval method using synthesized images from different camera views based on pixel-NeRF. The proposed method aims to retrieve images from various viewpoints as related images. To achieve this, we synthesize images whose camera views are different from the original query image as additional new queries for the retrieval. Then we determine the weight of the synthesized images that indicates how much they contribute to the retrieval. The proposed method can effectively utilize synthesized images for the retrieval and provide desired images independently from the camera view of the query image. Experimental results show that our method can improve the retrieval performance by utilizing effective synthesized images from different camera views.