Remote Sensing Image Retrieval with Gabor-CA-ResNet and Split-Based Deep Feature Transform Network

Abstract

In recent years, the amount of remote sensing imagery data has increased exponentially. The ability to quickly and effectively find the required images from massive remote sensing archives is the key to the organization, management, and sharing of remote sensing image information. This paper proposes a high-resolution remote sensing image retrieval method with Gabor-CA-ResNet and a split-based deep feature transform network. The main contributions include two points. (1) For the complex texture, diverse scales, and special viewing angles of remote sensing images, A Gabor-CA-ResNet network taking ResNet as the backbone network is proposed by using Gabor to represent the spatial-frequency structure of images, channel attention (CA) mechanism to obtain stronger representative and discriminative deep features. (2) A split-based deep feature transform network is designed to divide the features extracted by the Gabor-CA-ResNet network into several segments and transform them separately for reducing the dimensionality and the storage space of deep features significantly. The experimental results on UCM, WHU-RS, RSSCN7, and AID datasets show that, compared with the state-of-the-art methods, our method can obtain competitive performance, especially for remote sensing images with rare targets and complex textures.

Satellite Imagery Retrieval Based on Adaptive Gaussian Markov Random Field Model with Bayes Deep Convolutional Neural Network

“…The experimental results show no remarkable difference between individual datasets (with and without labels) and the combined dataset (with and without labels). GoogLeNet (Napoletano, 2018;Zhou et al, 2018), and LDCNN (Zhou et al, 2017), Gabor-ResNet, and Gabor-CA-ResNet (Zhuo and Zhou, 2021)-and that has been presented in Table 5. This paper has taken the results obtained by other networks for comparison purposes because it is challenging to implement all the networks in a PC system.…”

“…Therefore, to validate the AGMRF-BDCNN method's performance, this study has taken into account only the benchmark state-of-the-art methods, such as LDCNN, VGGS_Fc2, GoogLeNet, ResNet50, ResNet101, and SatResNet50. Zhuo and Zhou (2021) have introduced a Gabor-CA-ResNet model with modification of the ResNet by adding the Gabor filter and the Channel Attention mechanism. Further, they have introduced a split-based deep feature transform network that transforms the features from Gabor-CA-ResNet.…”

“…Literature (Napoletano, 2018;Justus et al, 2018;Dong and Luo 2020) reveal that the computational time is fully dependent on computer architecture and other factors because the running time of a 1,;2; 3, 4 : Napoletano, 2018;Zhou et al, 2018;Zhou et al, 2017;Zhuo and Zhou, 2021. particular method may vary from computer to computer accordingly the computer architecture. This problem can be avoided by running all the state-of-the-art approaches and the proposed method in a single PC system.…”

“…Imagery retrieval assists in exploring the same or similar imageries from the repository. In the last decade, a significant effort has been made for HRRS imagery retrieval (Yang and Newsam, 2013;Blaschke et al, 2014;Du et al, 2016;Guo et al, 2019;Xiong et al, 2020;Zhuo and Zhou, 2021). The imagery retrieval is performed in two stages: (1) feature extraction, (2) matching and retrieval.…”

Image Retrieval Model Analysis of Digital Library Based on Texture Characteristics

“…Gabor filtering belongs to the spatial domain method and wavelet transform belongs to the frequency domain method. Zhuo and Zhou [15] proposed a texture feature extraction algorithm based on Gabor filtering. Li et al [16] overcame the deficiency of the feature extraction algorithm based on frequency domain and proposed a texture feature extraction algorithm based on wavelet.…”