Ground T-shaped moving target detection and positioning method based on monocular vision
A monocular vision and moving target technology, applied in image data processing, instruments, character and pattern recognition, etc., can solve problems such as large amount of calculation and the influence of color feature segmentation
Inactive Publication Date: 2018-11-06
HARBIN INST OF TECH
View PDF3 Cites 3 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Color images contain rich color information, are easy to apply, and have good real-time performance. However, changes in various factors such as illumination, shadows, and camera characteristics will affect the segmentation of color features.
The edge is a high-level feature obtained from the comprehensive processing of multiple pixels in the image, which can well describe the contour information of the object itself, and the edge-based target detection algorithm has good accuracy and robustness, but the amount of calculation is relatively large. Big
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 Construction
[0071] The steps of a method for detecting and locating a ground T-shaped moving target based on monocular vision described in this embodiment are as follows:
[0072] Step 1. Collect a certain number of positive samples and negative samples, extract HOG features from the collected samples, and assign labels, send the features and labels to SVM (Support Vector Machine) for training, and save the trained classifier;
[0073] Step 2. Input the picture collected by the camera on the drone. The camera looks straight down and is connected to the drone, so we can get the height of the camera from the ground. At the same time, assuming that the height of the target from the ground is known, the camera can be obtained distance to target. And the UAV attitude change range is small, therefore, it can be approximated that the camera is looking straight down. The size l in the actual space of the target x *l y , is known, and the height h of the camera is known, and we use a fixed foca...
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 provides a ground T-shaped moving target detection and positioning method based on monocular vision, relates to the technical field of target detection, and aims to realize accurate acquisition of the position and the direction of a T-shaped board on the ground on a rotor unmanned aerial vehicle. The method includes steps: firstly acquiring a certain quantity of target positive sample pictures and negative sample pictures, and generating a classifier according to the pictures; then inputting a picture acquired by a camera, obtaining different regions of the picture through a detection window of a certain size, extracting HOG characteristics of the obtained regions, and inputting the characteristics to the trained classifier to determine whether the regions are the target; forthe target region, performing color space transformation, performing image segmentation according to a threshold, generating a binary image, and conducting a shape analysis on the binary image to obtain the direction of the T-shaped board; and finally, performing coordinate transformation according to attitude and height information of the camera to obtain the relative position and direction of the target and an aircraft. The method is used for obtaining the relative position and the moving direction of the ground moving target by the unmanned aerial vehicle.
Description
technical field [0001] The invention relates to a method for detecting and locating a ground target on a rotor unmanned aerial vehicle, and relates to the technical field of target detection. Background technique [0002] The rotor unmanned aerial vehicle is a vertical take-off and landing unmanned aerial vehicle with novel appearance and superior performance. It has the characteristics of simple structure, flexible operation and strong carrying capacity. Wide application value. [0003] For UAVs, the detection and positioning of ground targets is the basis for realizing the interaction between UAVs and the environment, which can improve the automation and intelligence level of flight control. During the mission phase of UAVs, the detection and positioning of targets can improve the ability of UAVs to perform tasks autonomously. [0004] For the target detection problem, there are already some techniques. There are mainly two types of target detection methods. One is the ...
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 IPC IPC(8): G06T7/136G06T7/246G06T7/70G06K9/62
CPCG06T7/136G06T7/246G06T7/70G06F18/2411
Inventor 侯谊贺风华姚郁马杰郝宁
Owner HARBIN INST OF TECH