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

Bokeh method based on salient region detection model

A technology of region detection and background blur, applied in biological neural network model, image data processing, image enhancement and other directions, can solve problems such as unclear boundaries, and achieve the effect of clear salient boundaries

Inactive Publication Date: 2018-06-29
FUZHOU UNIV
View PDF5 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] Aiming at the problems of unclear boundaries in the existing background blurring methods, the present invention proposes a background blurring method based on a salient area detection model, which can detect the entire salient area, and can detect objects including multiple salient objects and small-scale salient objects. It performs well in various complex situations such as sexual objects, not only can accurately detect the complete salient area, but also the salient boundary is relatively clear

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
  • Bokeh method based on salient region detection model
  • Bokeh method based on salient region detection model
  • Bokeh method based on salient region detection model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0060] The present invention will be further described below in conjunction with the drawings and embodiments.

[0061] Such as figure 1 As shown, the present invention provides a background blurring method based on a saliency area detection model, which includes the following steps:

[0062] Step S1: Obtain the original image;

[0063] Step S2: Construct a saliency region detection model based on the convolutional neural network to obtain the saliency map of the original image;

[0064] Step S3: Put the saliency map into the fully connected conditional random field for training, and obtain the optimized saliency map;

[0065] Step S4: Binarize or segment the optimized saliency map to obtain 01 matrix SBM, obtain foreground index matrix IF and background index matrix IB, which are defined as follows:

[0066]

[0067] Among them, M×N is a matrix of all ones with the same resolution as the original image;

[0068] Step S5: Use the distance weighted average algorithm to realize the global b...

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 bokeh method based on a salient region detection model. The method comprises the following steps: obtaining an original image, constructing a salient region detection model on the basis of a convolutional network to obtain a salient image of the original image, training the obtained salient image in a fully connected conditional random field to obtain an optimized salientimage, performing binaryzation or segmentation processing on the optimized salient image to obtain a 01 matrix, and obtaining a foreground index matrix and a background index matrix; realizing globalblurring of the original image with a distance weighted average algorithm; finally, splicing an original foreground image with the blurred background image to obtain the blurred background image. With adoption of the method, not only can a complete salient region be detected accurately, but also the salient boundary is relatively clear, so that features of the foreground image can be kept duringbokeh, and content of the foreground image is not damaged.

Description

Technical field [0001] The invention relates to the technical field of digital image processing, in particular to a method for background blurring based on a salient region detection model. Background technique [0002] Image background blur is a very common processing process in tasks such as image rendering, beautification, and enhancement. It can effectively highlight target objects and dilute the background information, thereby enhancing the visual effect. At present, some image processing software performs this processing well, but its processing methods all require manual annotation of the foreground area, which requires a lot of manpower and is not convenient for mass processing; in addition, the existing technology's fuzzy diffusion methods are all It is a regular shape and it is difficult to adapt to complex and changeable image content. The existing automatic background blur technology is immature in the foreground edge extraction, resulting in unclear boundaries, cutt...

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
IPC IPC(8): G06T3/00G06T7/11G06N3/04
CPCG06T7/11G06T2207/10004G06N3/045G06T3/04
Inventor 余春艳徐小丹陈立杨素琼王秀
Owner FUZHOU UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com