Foreground extraction method based on double-frame encoding and decoding model

A technology of encoding and decoding and foreground extraction, which is applied in the field of foreground extraction based on a double-frame encoding and decoding model, can solve the problems of inability to share basic image features and reduce model running speed, and achieve superior convenience, ensure real-time performance, and facilitate deployment Effect

Inactive Publication Date: 2020-02-28
中国石化销售股份有限公司 +1
View PDF10 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Directly combining optical flow prediction cannot achieve the sharing of basic image features, which will lead to a decrease in model running speed

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
  • Foreground extraction method based on double-frame encoding and decoding model
  • Foreground extraction method based on double-frame encoding and decoding model
  • Foreground extraction method based on double-frame encoding and decoding model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] In order to enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings, but this does not constitute a limitation to the present invention.

[0020] A kind of foreground extraction method based on double-frame encoding and decoding model that the present invention proposes, its steps are:

[0021] 1) Preprocessing the input two frames of images, including noise removal and normalization. Noise was removed by median filtering with a filter window of 5. The normalized formula is as follows:

[0022]

[0023] Where x is the pixel value in the value range [0,255], which is divided into three channels of RGB.

[0024] 2) Use the basic skeleton of the classification network to perform feature extraction on the input image respectively, and generate feature maps of different scales. From the first level to the sixth level, the sizes are 1 / 1, 1 / 2,...

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 foreground extraction method based on a double-frame encoding and decoding model, which belongs to the field of computer vision and comprises the following steps: training the double-frame encoding and decoding model, and performing preprocessing, feature extraction, feature map fusion, convolution and up-sampling on a double-frame image after training until a final fusion feature map with the same size as an input image is obtained; and performing softmax on the final fusion feature map to obtain a corresponding foreground prediction probability, and performing binarization to extract a foreground region. According to the method, corresponding image features can be generated while the motion information is generated, and finally prediction of the foreground target is realized.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a foreground extraction method based on a two-frame encoding and decoding model. Background technique [0002] With the popularity of computers and cameras, more and more video recordings need to be automatically analyzed. The method of manual analysis not only consumes a lot of labor costs, but also is difficult to be reliable and real-time. Using computer to process video is the main development direction of video processing at present. Foreground extraction is one of the basic technologies for computer processing video. It can focus video research and analysis on a small number of foreground target areas, and perceive important areas in video images, which can be used to help users browse important information in videos. Improve the efficiency of users analyzing video recordings. [0003] A common algorithm for foreground extraction is the frame difference method,...

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 Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06K9/38G06K9/40G06N3/04G06N3/08
CPCG06N3/08G06V20/46G06V10/28G06V10/30G06N3/045G06F18/253
Inventor 赖少川杨大慎何永君陈晓华姜红涛王佳欣马翠霞
Owner 中国石化销售股份有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap