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

Method for changing automobile body color in automobile image and algorithm structure thereof

An algorithm structure and image technology, applied in the field of image processing, can solve problems such as low availability rate, unsatisfactory effect, uneven color, etc.

Active Publication Date: 2019-10-01
BEIJING BAIDU NETCOM SCI & TECH CO LTD
View PDF6 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The car color change pictures generated by the traditional Generative Adversarial Networks (GAN, Generative Adversarial Networks) model have low availability, uneven colors and unsatisfactory effects

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
  • Method for changing automobile body color in automobile image and algorithm structure thereof
  • Method for changing automobile body color in automobile image and algorithm structure thereof
  • Method for changing automobile body color in automobile image and algorithm structure thereof

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0095] An embodiment of the present invention provides a method for modifying a local area of ​​an image, the method comprising: S1) acquiring a first image and a pixel vector, and determining a local area in the first image; S2) using the pixel vector in the local Intra-region mapping generates local regions having pixel features corresponding to said pixel vectors;

[0096] Such as figure 1 , the local area can be set as the position of the vehicle body, and the first image can be set as having the background BG( figure 1 box with two dotted four-pointed stars) and an image of a car with body color a x a ( figure 1 The area with slash in is the front windshield or window glass), and the pixel feature of the pixel vector is the color feature (let it be color b).

[0097] Specifically, determining the local area in the first image in step S1) includes:

[0098] S101) Determine an object in the first image;

[0099] S102) Perform edge detection and image segmentation on th...

Embodiment 2

[0128] Such as figure 1 , an algorithm structure for changing a local area of ​​an image, the algorithm structure includes: a generative confrontation network, which is used to map the input first image and the input pixel vector in the local area of ​​the first image to generate a The local area of ​​the pixel feature corresponding to the pixel vector.

[0129] Optionally, the generative model of the generative confrontation network includes: an encoder, a hidden layer space vector, and a decoder; the input layer of the encoder is used to receive the data of the first image; the hidden layer space The vector is used to transfer the data output by the output layer of the encoder to the input layer of the decoder; the hidden layer of the decoder is used to receive the data output by the input layer of the decoder and collect and fuse the Data of the hidden layer of the encoder; wherein, the hidden layer of the decoder is also used to receive the pixel vector and output a local...

Embodiment 3

[0144] Such as Figure 5 , a training method for changing an image local area algorithm, the training method includes:

[0145] S1) Obtain a set of labeled sample data, wherein the set of labeled sample data includes different images of objects with different pixel characteristics, and the local area of ​​the image where the object is located and corresponding to the object has a labeled mask area and also has a The labeled pixel vector corresponding to the pixel feature;

[0146] S2) Build a generative confrontation network, the generative confrontation network includes a generative model and a discriminant model, and the generative model has the characteristics of updating historical output data to current input data and obtaining partial data consistent with the characteristics of historical input data The loop function of the current output and the loop function is used to construct the loop consistency loss and the reconstruction loss, the generative model also has a dec...

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 provides a method for changing an image local area, and belongs to the technical field of image processing. The method comprises: acquiring a first image and a pixel vector, and determining a local area in the first image; and mapping in the local area by using the pixel vector to generate a local area with a pixel feature corresponding to the pixel vector. According to the method, the automobile picture with the real automobile body color can be generated, the diversity of advertisement materials is improved, the color changing accuracy of the model is improved, the automobile body segmentation capability is further achieved, and the color changing uniformity is improved in the color changing process of the model.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method for changing a local area of ​​an image, an algorithm structure for changing a local area of ​​an image, a training method for changing an algorithm for a local area of ​​an image, a method for changing a local area of ​​an image A device for image local area and a computer readable storage medium. Background technique [0002] With the rapid development of the Internet, Internet advertising has become one of the most important forms of advertising today. Due to its fast, efficient, and ideal effect, Internet advertising is favored by various small and medium-sized enterprise customers. In order to increase the click-through rate of advertisements, the styles of Internet advertisements are also emerging in an endless stream, specifically including text, pictures, videos and other forms. The richness of advertisement content and the diversity of advertise...

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): G06T11/00G06K9/62G06N3/04G06N3/08G06T7/13G06T7/194
CPCG06T11/001G06T7/13G06T7/194G06N3/08G06T2207/10024G06N3/045G06F18/2411
Inventor 朱曼瑜刘霄文石磊孙昊张赫男
Owner BEIJING BAIDU NETCOM SCI & TECH CO LTD
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