17530203. REDUCED BANDWIDTH IMAGE TRANSFER simplified abstract (Capital One Services, LLC)
Contents
REDUCED BANDWIDTH IMAGE TRANSFER
Organization Name
Inventor(s)
Austin Walters of Champaign IL (US)
Vincent Pham of Seattle WA (US)
Jeremy Goodsitt of Champaign IL (US)
REDUCED BANDWIDTH IMAGE TRANSFER - A simplified explanation of the abstract
This abstract first appeared for US patent application 17530203 titled 'REDUCED BANDWIDTH IMAGE TRANSFER
Simplified Explanation
The patent application describes a method to reduce bandwidth for image transfer by converting images to a color space with a luminance channel and removing other color channels. This allows for transmission of the image with less bandwidth compared to the original colorized image.
- Image transfer bandwidth is reduced through color channel reduction
- Images are converted to a color space with a luminance channel
- Color channels except for the luminance channel are removed from the image
- Transmission of the image requires less bandwidth than the original colorized image
- A model, such as a convolutional neural network, can be generated and shared over the network
- The model can colorize an image with missing channel information
- The decolorized image is received and the model is invoked to re-colorize it based on luminance data and color hints
- The model can produce custom colorization for display or user preferences
Potential Applications
- Image transfer and sharing platforms with limited bandwidth
- Remote viewing of images with reduced bandwidth requirements
- Mobile applications where bandwidth is limited or expensive
Problems Solved
- High bandwidth requirements for image transfer
- Limited bandwidth availability in certain networks or situations
- Inefficient use of bandwidth for colorized images
Benefits
- Reduced bandwidth requirements for image transfer
- Faster transmission of images over networks
- Improved accessibility to images in low-bandwidth environments
- Customizable colorization options for users or displays
Original Abstract Submitted
Bandwidth for image transfer is reduced through color channel reduction. An image can be converted to a color space with a luminance channel, and the image can be decolorized by removing channels except for the luminance channel. Transmission of the image can subsequently be initiated across a network utilizing less bandwidth than the colorized image. Further, a model, such as a convolutional neural network, can be generated and shared over the network. The model is operable to colorize an image with missing channel information. After receiving the decolorized image, the model can be invoked to re-colorize the image based on luminance data and any color hints provided. The model can also produce a custom colorization for a display, a user, or both.