Light Invariant Video Imaging (LIVI)

Shadow Free Objects Which Appear the Same Independent of Light Conditions

 Imaging Capabilities / Applications / How It Works

 

Concept

Removing the Effect of the Background Lights

LIVI removes the effect the background light have on the scene and leave only the flash light.

Imaging Capabilities

LIVI robustness to different light conditions results from its unique imaging capabilities. The system output frames are not affected by ambient sources (like the Sun) and direct sources (like a flash light). Objects captured by the system are:

 
 

Shadow Free Video Frames

Shadows from the flash light are created are unseen by the camera because they are situated behind the objects. In this case all the shadows appearing in the frame originate from the backgrround lights. Removing the effect of the background lights will also remove the shadows.

 

Color Constancy

Objects captured by LIVI appear to illuminate only by the flash light. This light souce predectability creates ideal conditions for any color constancy or white balance algorithm. This results in objects appearing to have the same color under any light condition.

 

Contrast Enhancement

Objects captured by LIVI appear to illuminate only by the flash light. These uniform light conditions creates a contraced enhanced video frames.

 

Applications

LIVI delivers fixed object appearance under any light conditions. This can improve the performance of computer vision applications such as face recognition and object detection and classification.

Improved Face Recognition

Faces captured by LIVI appear to illuminate only by the flash light. These constant light conditions creates ideal scenario and improved the performance of face detection and recognition algorithms.

Read paper

 

Better Object Detection and Classification

Objects captured by LIVI have constant light conditions - which are similar to the illumination used in the training data. These constant light conditions creates ideal scenario and improved the performance of object detection and classification. Using LIVI as a feedback in the training session can further improve CNN performance even when not using it as the input.

Read Paper

 
 

How It Works

PrototypeWorking Principle

Prototype

LIVI prototype is built from readily available, off-the-shelf components.

  • USB camera

  • Controlled LED light system comprised of:

    • LED light.

    • LED driver.

    • Programmable PWM controller.

There is no synchronization between light and camera. A product based on our prototype's working principle could fit into a laptop, tablet or cellphone.

 

Working Principle

LIVI is based on principles from communication systems such as Amplitude Modulated (AM) radio. Light intensity is modulated in time and the camera samples N frames and outputs a single frames using linear signal reconstruction. 

 
 

More Details

For an in-depth description of the technology behind LIVI, please refer to our paper and accompanying video.

 

Amir Kolaman, Maxim Levov, Rami Hagege  and Hugo Guterman, " Amplitude Modulated Video Camera - Light Separation in Dynamic Scenes",  IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016

Further detail on application of the method for improving object and classification in neural networks, please refer to our paper:

"Light Invariant Video Imaging for Improved Performance of Convolution Neural Networks", IEEE Transaction on Circuits and Systems for Video Technology

Patent Pending

Press Release

28/02/2018

LIVICNNPaper.jpg

Department of Electrical and Computer Engineering

Ben Gurion University of the NEGEV

© 2017 Amir Kolaman

This site was designed with the
.com
website builder. Create your website today.
Start Now