Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Vision Detection Method for Planar Rotating Targets

A visual detection and plane rotation technology, applied in the fields of computer vision, pattern recognition and machine learning, can solve problems such as increasing the number of false positive results, reducing detection accuracy, and limiting detection efficiency, so as to reduce computing expenses, improve detection efficiency, and avoid The effect of feature extraction and classification judgment

Active Publication Date: 2019-05-31
SANITARY EQUIP INST ACAD OF MILITARY MEDICAL SCI PLA
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The above method needs to perform feature extraction and classification judgment in multiple directions on the detection window at each position of the image to be tested. The huge amount of calculation limits the detection efficiency to a certain extent; and repeated detection at each position will also increase the probability of false positive results. Quantity, reduce detection accuracy

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Vision Detection Method for Planar Rotating Targets
  • A Vision Detection Method for Planar Rotating Targets
  • A Vision Detection Method for Planar Rotating Targets

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0039] An embodiment of the present invention provides a visual detection method for planar rotating targets, see figure 1 , the method includes the following steps:

[0040] 101: Randomly rotate the training samples, obtain target sample sets with various orientations, combine the negative sample sets that do not contain the target, obtain the first image features after image feature extraction, use the first image features as input, and pass the machine The learning method performs binary classification training to obtain the primary screening classifier;

[0041] Wherein, the above-mentioned machine learning method may be any machine learning method that can realize classification, such as support vector machine, random forest algorithm, neural network, deep learning, etc., which is not limited in the embodiment of the present invention.

[0042] Further, the above-mentioned steps of binary classification training are well known to those skilled in the art, and are not des...

Embodiment 2

[0053] Combine below Figure 2-Figure 4 The scheme in embodiment 1 is introduced in detail, see the following description for details:

[0054] 201: Train the primary screening classifier;

[0055] In the detection task, the target often only occupies a small part of the picture to be tested, and the detailed multi-directional detection for each position in the picture to be tested will undoubtedly increase unnecessary computing expenses. Therefore, the embodiment of the present invention applies a window preliminary screening link at the beginning of the detection, that is, scans and detects the entire image to be tested without considering the direction, and obtains candidate detection windows for suspected objects.

[0056] Before the training starts, it is necessary to randomly rotate the training samples containing the target to obtain a positive sample set containing samples from all directions, such as figure 2 As shown, it illustrates some positive sample pictures use...

Embodiment 3

[0074] Combine below Figure 5 The scheme in embodiment 1 and 2 is carried out feasibility verification, see the following description for details:

[0075] exist Figure 5 In the shown experiments, the embodiment of the present invention uses the histogram of sector-shaped oriented gradients (SRHOG) as the image feature required for detection, and the classification method adopts the random fern algorithm (RFs). The training and test pictures are from the Freestylemotocross public dataset, the purpose is to detect motorcycles with different rotations in various pictures. In the picture to be tested (such as Figure 5 As shown in (a), when performing target detection, according to the above method, firstly, the candidate detection window set in the picture to be tested is obtained through window preliminary screening (such as Figure 5 (b), Figure 5 Each circular area in (b) represents a candidate detection window); then use the direction estimation link to estimate the d...

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

No PUM Login to View More

Abstract

The invention discloses a visual detection method for a plane rotating target, which relates to the fields of computer vision, pattern recognition and machine learning. The method can quickly and effectively solve the detection requirement for the rotating target in real situations. This method follows the visual detection framework that combines image feature expression and machine learning algorithms, but does not involve specific image features and machine learning methods, so it has good promotion value. The method divides the detection process into three steps: window screening, direction estimation and final verification. Specifically, window screening is to quickly and roughly scan the image to be tested, filter background information, and obtain candidate detection windows containing false positive results; direction estimation is to predict the direction of the assumed target in the candidate detection window; final verification is based on The obtained estimated direction is further judged on the candidate detection window. The method can reduce the calculation expenditure of the traditional multi-directional detection method, and greatly improve the detection efficiency of the rotating target.

Description

technical field [0001] The invention relates to the fields of computer vision, pattern recognition and machine learning, in particular to a visual detection method for a plane rotating target. Background technique [0002] As an important branch of computer vision, object detection is a fundamental problem for high-level image understanding and video analysis, and rotating object detection is a key component of this problem. For the detection of rotating targets, the most commonly used method is multi-directional detection, that is, rotating the image to be tested at a certain angle for multiple detections. [0003] In the process of realizing the present invention, the inventor finds that at least the following disadvantages and deficiencies exist in the prior art: [0004] The above method needs to perform feature extraction and classification judgment in multiple directions on the detection window at each position of the image to be tested. The huge amount of calculation...

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
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62
CPCG06F18/285G06F18/2411G06F18/214
Inventor 刘保真孙景工吴航苏卫华张文昌秦晓丽苑英海安慰宁
Owner SANITARY EQUIP INST ACAD OF MILITARY MEDICAL SCI PLA
Features
  • Generate Ideas
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com