A Novel Adaptive Feature Fusion Strategy for Image Retrieval

Lu, Xiaojun; Zhang, Libo; Niu, Lei; Chen, Qing; Wang, Jianping

doi:10.3390/e23121670

Cited by 7 publications

(14 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The grey-scale 3D topography and image information entropy 14,15 are used to analyse the characteristics of different areas and their impact on monitoring accuracy. Information entropy is generally used to quantify the degree of information confusion, and can also be used to count the expected value of random variables, so image information entropy is chosen as a metric to measure the richness of grey-scale information in different areas of the slope image.…”

Section: Analysis Of the Influencing Factors Of Regional Features By ...mentioning

confidence: 99%

Research on regionalised digital image slope monitoring methods

et al. 2023

View full text Add to dashboard Cite

The digital image correlation method is a high precision and non-contact ground deformation monitoring method, which is widely used in surface deformation monitoring. However, according to its principle, different surface characteristics have great influence on the monitoring results. In this paper, a regional digital image slope monitoring method is established to meet the monitoring requirements of different areas of the same slope. Firstly, the features of different regions in the image are identified according to the optical region detection, and then the deformation calculation method is matched according to the features of different regions. Finally, according to the binocular vision principle, the three-dimensional displacement of the slope is obtained. Taking a highway slope in Qinhuangdao as an example, we realised all-weather and full-field deformation monitoring of slope, and verified the applicability and feasibility of regional digital image slope monitoring method. The applicability of digital image correlation method in slope monitoring is improved.

show abstract

Section: Analysis Of the Influencing Factors Of Regional Features By ...mentioning

confidence: 99%

Research on regionalised digital image slope monitoring methods

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Lu et al [18] proposed a feature-fusion method based on information entropy theory and relevant feedback to adapt the weights of image features. However, their relevant feedback of a feature of interest is based on how good the feature is in selecting images that have the same category as the query image.…”

Section: B Deep Feature Similaritymentioning

confidence: 99%

Improving Deep-Feature Image Similarity Calculation: A Case Study on an Ukiyo-e Card Matching Game Lottery

2022

View full text Add to dashboard Cite

The purpose of this study is to improve AI players in Lottery. Lottery is a card matching game designed based on the concept of Audience Participation Game With a Purpose. Namely, it lets live streaming game audiences take part in gameplay and collects image similarity data perceived by these audiences. The game employs two AI players that initially calculate image similarities based on deep features (deep-feature similarities). In our previous study, it was found that similarities, between a certain pair of images, perceived by machines or computers -calculated based on deep features -were different from similarities perceived by humans. This would make gameplay by the AI players unbelievable, in other words, non-human-like. This study, therefore, proposes to use a linear model, built based on pre-collected human data, for improving the deep-feature similarities. The amount of human data required to make the model stable is also discussed. Experimental results show that the linear model only requires small quantities of human data to greatly improve the deep-feature similarities. At the same time, our results also show that the game Lottery is indispensable. This is because the linear model can only make the calculated similarity closer to that of humans, but there is still discernible difference; in order to obtain accurate similarity between images, it is necessary to collect a certain amount of human-perceived similarity data.INDEX TERMS Audience participation game, game with a purpose, audience participation game with a purpose, technique for similarity data collection, card matching game, human-perceived similarity, deep feature similarity.

show abstract

“…Image retrieval is an important computer vision task aimed at retrieving similar or related images based on user provided query images [1][2][3][4][5][6][7][8][9][10][11][12][13] . In the past decades, researchers have proposed many image retrieval methods.…”

Section: Introductionmentioning

confidence: 99%

“…SIFT is used for image analysis based on different orientations and scales. In [4][5][6][7][8] , the LBP and Color are both involved. In [9][10][11][12] , the author combines SIFT and color, provides a powerful image descriptor.…”

Section: Introductionmentioning

confidence: 99%

“…With the remarkable success of CNN networks in various practical applications, there has been an abundance of methods for image retrieval based on CNN features. In [4,8,[13][14][15][16] , the author incorporates CNN features for fusion has significantly improved retrieval accuracy. By integrating multiple feature information, it is possible to better capture the semantic content, structural information, and contextual relationships of images, thereby improving the accuracy and robustness of image retrieval [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An image retrieval method based on fusion of multiple features

Wang,

Dou,

Guo

et al. 2023

Third International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2023)

View full text Add to dashboard Cite

In image retrieval tasks, two main indicators are focused on: accuracy and efficiency. In general, the more the number of features, the higher the retrieval accuracy but the lower efficiency. Focusing on ensuring retrieval efficiency while improving retrieval accuracy, this paper designs an unsupervised image retrieval system that integrates multiple features. On UKbench, Holidays, UC Merced Land Use, and RSSCN7, the method proposed was validated. On UKbench, compared to methods based on shuffleNetV2, VGG-16 and Resnet18, the method proposed improves N-S by 0.1882, 0.2499, and 0.1985 respectively, while saving retrieval time by 0.0115s. On Holidays, compared to methods based on shuffleNetV2, VGG-16 and Resnet18, the method proposed improves MAP by 7.11%, 9.05%, and 7.27% respectively, while saving retrieval time by 0.0021s. On UC Merced Land Use, compared to methods based on shuffleNetV2, VGG-16 and Resnet18, the method proposed improves accuracy by an average of 3.31%, 11.84%, and 6.29% respectively, while saving retrieval time by 0.0007s. On RSSCN7, compared to methods based on shuffleNetV2, VGG-16 and Resnet18, the method proposed improves accuracy by an average of 2.19%, 9.19%, and 7.18% respectively, while saving retrieval time by 0.0002s. The experimental results show that the proposed method can effectively improve retrieval accuracy and to some extent improve retrieval efficiency.

show abstract

A Novel Adaptive Feature Fusion Strategy for Image Retrieval

Cited by 7 publications

References 25 publications

Research on regionalised digital image slope monitoring methods

Research on regionalised digital image slope monitoring methods

Improving Deep-Feature Image Similarity Calculation: A Case Study on an Ukiyo-e Card Matching Game Lottery

An image retrieval method based on fusion of multiple features

Contact Info

Product

Resources

About