Unlock instant, AI-driven research and patent intelligence for your innovation.

Fast and efficient pedestrian detection method

A pedestrian detection and efficient technology, applied in the field of image processing, can solve the problem of time-consuming detection process

Active Publication Date: 2021-11-19
北京智峰真道科技发展有限公司
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The BING feature scales the image, then calculates the gradient of the scaled image, and uses the trained BING template to convolve each pixel of the gradient image. When the 8×8 template convolutes each pixel 64 multiplications and 63 additions are required, which is a very time-consuming operation for the detection process

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
  • Fast and efficient pedestrian detection method
  • Fast and efficient pedestrian detection method
  • Fast and efficient pedestrian detection method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0053] Embodiments of the present invention: a fast and efficient pedestrian detection method, including a training phase and a detection phase; in the training phase:

[0054] 1) Create a data set, extract features from the created data set, train the SVM pedestrian detection model, and obtain the SVW discriminant model;

[0055] 2) Create a data set, use the K-Means++ algorithm to optimize the training samples, extract HWEBING features, train the HWEBING similarity model, and obtain the HWEBING detection model;

[0056] 3) Input the detection image, load the HWEBING detection model and the SVW discrimination model;

[0057] In the detection stage: 1) Obtain object-like windows by extracting HWEBING features, and use non-maximum value suppression to reduce the number of windows;

[0058] 2) Extract the HOG and MLBP features of the candidate pedestrian window;

[0059] 3) The SVM classifier is used for classification decision-making, and the detection result is obtained.

...

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 fast and efficient pedestrian detection method. The present invention makes improvements on the traditional pedestrian detection framework. Before extracting features, an improved algorithm based on BING feature HWEBING is proposed to pre-detect images. Pre-detection can reduce a large number of non-object windows and screen out possible Candidate windows for objects. The detection rate of the MLBP feature of the present invention is respectively increased by 3.5% and 2.1% compared with the detection rates of the uniform mode LBP and the basic mode LBP. And compared with the traditional pedestrian detection method, this method uses the HWEBING algorithm to pre-detect the picture and then extracts the HOG and MLBP features, which is 5.5 times faster than directly extracting the HOG and MLBP features. The combination of HWEBING algorithm and HOG‑MLBP features achieves better pedestrian detection results.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a fast and efficient pedestrian detection method. Background technique [0002] The traditional pedestrian detection framework obtains a pedestrian detector by extracting features from the data set and then training it with a classifier. In the detection stage, the picture to be detected is used as the input of the detector, and finally the detection result is output. [0003] The BING feature scales the image, then calculates the gradient of the scaled image, and uses the trained BING template to convolve each pixel of the gradient image. When the 8×8 template convolutes each pixel 64 multiplications and 63 additions are required, which is a very time-consuming operation for the detection process. [0004] K-Means algorithm is a common clustering algorithm. It first randomly selects some initial points as cluster centers, then calculates the distance between each samp...

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): G06K9/00G06K9/62
CPCG06V40/103G06F18/23213G06F18/2411
Inventor 张永军秦永彬许尽染肖伶
Owner 北京智峰真道科技发展有限公司