Basement parking-oriented robust visual feature matching and detection method
A technology of visual features and detection methods, applied in the field of robust visual feature matching and detection for basement parking, to achieve high positioning and mapping work, improve illumination robustness, and reduce costs
Pending Publication Date: 2022-07-22
TONGJI UNIV
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0007] The purpose of the present invention is to provide a robust visual feature matching and detection method for basement parking in order to overcome the above-mentioned defects in the prior art. High-precision positioning and map building, and then assist the vehicle to complete automatic parking
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
[0099] like figure 1 As shown, a robust visual feature matching and detection method for basement parking includes the following steps:
[0100] S1. Collect the original RGB image to perform offline training on the feature point detection deep learning model, and obtain a trained feature point detection model;
[0101] In the present embodiment, the neural network model based on the UnsuperPoint method is adopted for offline training, and the specific process of offline training is:
[0102] S1.1: Preprocess the collected raw RGB images (such as figure 2 shown) to obtain paired images, where the original RGB images were collected by a monocular visible light camera, and the preprocessing includes the following steps:
[0103] S1.1.1: For input image I o Perform pre-transformation to generate the initial image I' o . Pre-transform operations include random horizontal flips, vertical flips, and random cropping. Random cropping refers to random scaling of the image and cro...
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 robust visual feature matching and detection method for basement parking, and the method comprises the steps: collecting an original RGB image, and carrying out the offline training of a feature point detection deep learning model, and obtaining a trained feature point detection model; inputting to-be-detected data into the feature point detection deep learning model, and performing feature point detection to obtain a corresponding feature point detection result; inputting the to-be-detected RGB image and the true value label into the storage location detection deep learning model for offline training to obtain a trained storage location detection model; inputting the RGB image to be detected into the garage location detection deep learning model, and carrying out garage location detection to obtain a corresponding garage location detection result; and matching the feature point detection result with the storage location detection result to obtain a matching result. Compared with the prior art, the method has the advantages that the feature points can be detected under the condition of dim illumination or unobvious texture, so that high-precision positioning and mapping are realized, and the vehicle is assisted to complete automatic parking.
Description
technical field [0001] The invention relates to the technical field of automatic driving, in particular to a robust visual feature matching and detection method for basement parking. Background technique [0002] With the development of autonomous driving technology, autonomous parking has become a current research hotspot. Automatic parking mainly uses sensors all over the vehicle itself and the surrounding environment to measure the relative distance, speed and angle between the vehicle itself and surrounding objects, and then calculates the operation process through the on-board computer platform or cloud computing platform, and controls the vehicle's Steering and acceleration and deceleration enable the vehicle to realize automatic parking, parking and partial driving functions. The automatic parking process can roughly include the following five links: environmental perception, parking space detection and recognition, parking path planning, parking path following, and ...
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): G06V20/58G06V10/82G06T3/00G06T15/20G06N3/04G06N3/08
CPCG06T15/20G06N3/04G06N3/088G06T2207/20081G06T2207/20084G06T2207/30264G06T3/04
Inventor 田炜文永琨初新宁
Owner TONGJI UNIV