Closed-loop detection method based on local and convolutional neural network features
A convolutional neural network and closed-loop detection technology, which is applied in biological neural network models, neural architectures, computer components, etc., to achieve the effect of strong generalization ability, small memory footprint, and good robustness
Pending Publication Date: 2021-02-19
GUANGDONG POWER GRID CORP ZHAOQING POWER SUPPLY BUREAU
View PDF3 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] In order to overcome the problem that the detection in the prior art is difficult to cope with strong appearance changes, the present invention provides a closed-loop detection method based on local and convolutional neural network features, which combines the local feature descriptor LDB with the global feature extracted by the convolutional neural network Combined, in the process of closed-loop detection, it can not only deal with the scene of image appearance change, but also obtain the geometric topology information between images, so as to obtain a higher recall rate
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0043] Such as Figure 1-2 Shown is an embodiment of a closed-loop detection method based on local and convolutional neural network features, including the following steps:
[0044] Step 1: The current input image I collected by the mobile robot i Perform preprocessing and resize the image to 224×224 pixels. The convolutional neural network VGG16 is used to extract the global features of the current input image. The VGG16 network is pre-trained by the Places365-standard dataset, and the output of the penultimate fully connected layer of the network will be used as the image I i The global feature f glo,i , the dimensionality of the image global feature is 4096. The extracted global features are gradually inserted into the hierarchical navigable small world map (HNSW) of the approximate nearest neighbor search algorithm;
[0045] Step 2: Within the retrieval range of the current input image, search through HNSW and the current input image I i The most similar image is used...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to a closed-loop detection method based on local and convolutional neural network features, and the method comprises the following steps: extracting global image features of a collected input image through employing a convolutional neural network, and inserting the extracted global features into a small world graph; in the retrieval range of the current input image, retrieving an image most similar to the current input image as a closed-loop candidate image of the current input image through HNSW; introducing geometric consistency check, and matching feature points of thetwo images; inputting the matched feature points of the two images into a random sampling consistency algorithm, and if the number of internal points between the two images is greater than a threshold value, determining that the two images may form a closed loop; introducing a time consistency test, and if two continuous frames of images after the current input image meet a threshold condition, considering that the input image and the closed-loop candidate image form a group of closed loops. In a closed-loop detection process, a scene of image appearance change can be processed, and geometrictopology information between images can be obtained.
Description
technical field [0001] The invention relates to the field of vision-based positioning and navigation in the autonomous inspection of UAVs, and more specifically, to a closed-loop detection method based on local and convolutional neural network features. Background technique [0002] During the autonomous inspection process of UAVs, UAVs need to independently determine the required operations based on environmental information. Therefore, autonomous positioning and environmental map perception and construction are the key links in the autonomous inspection of UAVs. In recent years, with the improvement of computer hardware level and the development of visual processing technology, visual SLAM (simultaneous localization and mapping) technology has been widely used in mobile robot positioning and navigation tasks. Closed-loop detection is an important part of the visual SLAM system. Its main function is to judge whether the mobile robot has passed the place it has visited. It ...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/532G06F16/583G06K9/46G06N3/04
CPCG06F16/532G06F16/583G06V10/40G06N3/045
Inventor 游林辉胡峰孙仝陈政张谨立宋海龙黄达文王伟光梁铭聪黄志就何彧陈景尚谭子毅尤德柱区嘉亮罗鲜林
Owner GUANGDONG POWER GRID CORP ZHAOQING POWER SUPPLY BUREAU