A Robust Invariant Local Feature Matching Method for Changing Scenes

Abstract

The precise evaluation of camera position and orientation is a momentous procedure of most machine vision tasks, especially visual localization. Aiming at the shortcomings of local features of dealing with changing scenes and the problem of realizing a robust end-to-end network that worked from feature detection to matching, an invariant local feature matching method for changing scene image pairs is proposed, which is a network that integrates feature detection, descriptor constitution, and feature matching. In the feature point detection and descriptor construction stage, joint training is carried out based on a neural network. In the feature point extraction and descriptor construction stage, joint training is carried out based on a neural network. To obtain local features with solid robustness to viewpoint and illumination changes, the Vector of Locally Aggregated Descriptors based on Neural Network (NetVLAD) module is introduced to compute the degree of correlation of description vectors from one image to another counterpart. Then, to enhance the relationship between relevant local features of image pairs, the attentional graph neural network (AGNN) is introduced, and the Sinkhorn algorithm is used to match them; finally, the local feature matching results between image pairs are output. The experimental results show that, compared with the existed algorithms, the proposed method enhances the robustness of local features of varying sights, performs better in terms of homography estimation, matching precision, and recall, and when meeting the requirements of the visual localization system to the environment, the end-to-end network tasks can be realized.

A Visual Enhancement Network with Feature Fusion for Image Aesthetic Assessment

“…This indicates that SDFF module broadens the network structure of VGG16, deepening the memory of FF-VEN and reducing the number of required samples. As mentioned in [32], the score distribution of images with Mean in [0,4) or [7,10] tends to be Gamma. The number of those images account for 4.5% of all images.…”

“…For multi-scale image processing, Szegedy et al [21] proposed GoogLeNet, increasing the width of the network via sparse connections. Because of its great performance on ImageNet, they developed InceptionNet, using optimization algorithms to improve the performance of the model [7]. DMA-NET [22] performed random image clipping and extracted local fine-grained information.…”

“…The deep feature perceives abstract semantic information. Similar to InceptionNet [7], SDFF module broadens the network structure, aiming to improve the performance of models. We use VGG16 [10] as the baseline and take out the shallow feature and the deep feature from different convolution layers.…”

“…The contrast between them suggests the superiority of our network. We list three advanced methods: NIMA [8], ResNet [36], and InceptionNet [7]. In their experiments, the network on the basis of InceptionNet performed best, with Accuracy larger more than 2% than InceptionNet.…”

“…Neural networks have shown good advantages for image analysis and processing [4]- [6]. Dai et al [7] introduced the existing research in the field of intelligent media. The pooling layers was utilized to increase the speed of processing the low-level features [8].…”