A Step-by-Step Jointly Supervised Illumination Estimation Method
A lighting estimation and branching technology, which is applied in the field of lighting estimation with step-by-step joint supervision, can solve the problems of inaccurate lighting information, mismatching of lighting brightness and input image, mismatching of lighting information and texture information, etc., and achieve the goal of improving accuracy Effect
Active Publication Date: 2022-07-05
WUHAN UNIV
View PDF7 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] In order to overcome the defects of the prior art, the present invention provides a step-by-step joint supervision illumination estimation method, based on a single RGB image, a method of predicting the illumination information of any position in it, aiming to solve the problem of illumination estimation in the existing illumination estimation method. The information is not accurate enough, and the lighting information does not match the texture information, and the lighting brightness does not match the input image.
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0044] In order to further understand the content of the present invention and make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be understood that, the present invention is only explained and not limited herein.
[0045] like figure 1 As shown, it is the illumination estimation network of this embodiment. The camera rotation pose estimation network in this embodiment consists of an encoder, a decoder and a regression module, and the output camera rotation pose is expressed in the form of a quadruple. The panorama completion network of this embodiment includes a lighting decoder and a texture decoder, which respectively predict the exposure level map and the completed LDR panorama, and the two decoders share the same encoder. The lighting decoder is connected with the texture decoder through a dense attention m...
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
Login to View More Abstract
The invention designs a step-by-step joint-supervised illumination estimation method, which can estimate illumination information at any position in the picture by only using a single RGB picture. The original task is decomposed into three simpler subtasks of processing in-camera information, predicting out-of-camera information, and inverse mapping low dynamic range images to high dynamic images using the structure of a stepwise network. Use the vertical line segment to represent the predicted position to get a more accurate deformation algorithm, and use the exposure level map to directly supervise the position and intensity of the light to improve the accuracy of the light information. In addition, the lighting information is shared to the texture branch of the panorama through a dense lighting attention network, so that the generated texture information matches the lighting information. The present invention uses an intensity calibration algorithm to obtain an HDR panorama that matches the intensity of the input image. Lighting information can be estimated anywhere in the image using only a single RGB image.
Description
technical field [0001] The invention belongs to the field of image-based illumination estimation, and specifically refers to a step-by-step joint supervision illumination estimation method, which can estimate illumination information at any position in the picture by only using a single RGB picture. Background technique [0002] With the rapid development of computer technology, the field of computer vision has reached an unprecedented height. Illumination estimation can predict the illumination information of any location in a photo, which is a fundamental task of augmented reality. Today, augmented reality has been widely used in many fields, such as gaming, education, medical and military. Illumination estimation is the first step for these domains to perceive illumination information, and the accuracy of illumination prediction will directly affect the final generation results of these domains. The representation methods of illumination are mainly divided into two type...
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
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06T7/00G06T5/00G06N3/08
CPCG06T7/0002G06N3/08G06T2207/10024G06T2207/20081G06T2207/20084G06T5/92
Inventor 肖春霞于寒宁罗飞
Owner WUHAN UNIV