Persona Extraction via Depth Probability Maps

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional videoconferencing techniques often transmit irrelevant and distracting background information, reducing the quality and aesthetic of online meetings by not effectively isolating the presenter's image from the background.

Innovation Solution

The technology extracts a 'persona' image from video data, identifying and isolating the presenter using image depth data and probability maps, combining it with other video content to enhance the presentation experience.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If the camera transmits the entire field-of-view including background, then the video display shows complete scene information, but irrelevant and distracting background information reduces the quality and aesthetic of the videoconference

Engineering Contradiction:
Improvebackground informationVSAvoiddistracting information
Core Design Contradiction:
Loss of informationVSObject-affected harmful factors

Solution Approach 1:

The patent segments the video feed into distinct components: the presenter (foreground) and the background. By applying depth data and persona identification algorithms, the system separates these elements and selectively transmits only the relevant presenter portion, eliminating distracting background information while preserving essential presentation content.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts the presenter's image from the complete video feed using depth-based segmentation and persona identification. This extraction process isolates the relevant foreground subject (presenter) from the background, transmitting only the extracted persona to improve video quality and reduce visual distractions.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If multiple persona identification modules are used to improve accuracy, then the presenter identification precision increases, but the device complexity increases

Engineering Contradiction:
Improvepresenter identification accuracyVSAvoidnumber of processing modules
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines multiple persona identification modules (including depth-based modules, color-based modules, and gradient-based modules) into an integrated processing pipeline. By merging these modules and combining their probability maps through aggregation, the system achieves high identification accuracy while managing complexity through unified processing architecture.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent implements a universal persona identification system that uses multiple modules serving different functions (depth analysis, color analysis, gradient detection) but all contributing to the same goal of presenter identification. This multi-functional approach allows the system to handle various lighting conditions, backgrounds, and presenter characteristics through a single integrated framework.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS9414016B2System and methods for persona identification using combined probability maps
Publication Date: 2016.08.09 PERSONIFY INC
  • US9414016B2 patent drawing
  • US9414016B2 patent drawing
  • US9414016B2 patent drawing

AI summary

Disclosed herein are systems and methods for extracting person image data comprising: obtaining at least one frame of pixel data and corresponding image depth data; processing the at least one frame of pixel data and the image depth data with a plurality of persona identification modules to generate a corresponding plurality of persona probability maps; combining the plurality of persona probability maps to obtain an aggregate persona probability map; and generating a persona image by extracting pixels from the at least one frame of pixel data based on the aggregate persona probability map.