Hardware-Agnostic Advertisement Detection via Video Fingerprinting

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing advertisement detection systems fail to accurately identify advertisements across multiple channels due to hardware dependency and variations in signal quality and luminance, leading to inaccurate detection and loss of revenue.

Innovation Solution

A hardware-agnostic method that extracts audio and video fingerprints, normalizes frames, scales, trims, and generates digital signature values to detect advertisements, using both supervised and unsupervised approaches for robust detection across channels.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional supervised machine learning based advertisement detection is used, then advertisement detection capability is provided, but hardware dependency causes inaccurate detection across multiple channels

Engineering Contradiction:

Improveadvertisement detection accuracyVSAvoidhardware independence

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent applies parameter changes by normalizing video frames through histogram equalization to standardize luminance values, and by transforming video data into frequency domain using Fourier transform. These parameter transformations eliminate hardware-specific variations in signal quality and luminance, enabling consistent advertisement detection across different channels and set-top boxes.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces traditional mechanical/supervised learning approaches with a fingerprinting-based detection system. Instead of relying on supervised machine learning that is sensitive to hardware variations, the system uses audio and video fingerprinting with Fourier transform analysis to create hardware-agnostic advertisement identification, substituting the detection mechanism to achieve hardware independence.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Ease of manufacture

If different channels record programs in different contrast and brightness settings, then channel-specific optimization is achieved, but advertisement detection accuracy deteriorates

Engineering Contradiction:

Improvechannel-specific optimizationVSAvoidadvertisement detection precision

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent applies equipotentiality by equalizing the luminance histogram of video frames from different channels. Through histogram equalization, the system transforms frames with different contrast and brightness settings into a common luminance distribution, creating equipotential conditions that enable accurate advertisement detection across channels with varying recording settings.

Inventive Principle:

Principle #12Equipotentiality

3Adaptability or versatility

If set top boxes with different decoding quality are used, then user preference accommodation is achieved, but fingerprint consistency deteriorates

Engineering Contradiction:

Improveuser preference accommodationVSAvoidfingerprint consistency

Core Design Contradiction:

Adaptability or versatilityVSStability of the object's composition

Solution Approach 1:

The patent replaces reliance on consistent decoding quality with a robust fingerprinting system that uses Fourier transform analysis. This substitution makes the detection mechanism insensitive to variations in set-top box decoding quality, maintaining fingerprint consistency across different hardware configurations while accommodating user preferences for different decoding capabilities.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

4Measurement precision

If supervised machine learning approach is used for advertisement detection, then known advertisement matching is enabled, but new advertisement detection capability is lost

Engineering Contradiction:

Improveknown advertisement matching accuracyVSAvoidnew advertisement detection capability

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent applies universality by creating a fingerprinting system that serves multiple functions: it can match known advertisements through database comparison and simultaneously detect new advertisements through unsupervised fingerprint analysis. The Fourier transform-based fingerprint extraction provides a universal representation that works for both supervised matching and unsupervised detection of novel advertisements.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS10117000B2Method and system for hardware agnostic detection of television advertisements

Publication Date: 2018.10.30 SILVEREDGE TECH PVT LTD

AI summary

A system and method for hardware agnostic detection of one or more advertisements broadcasted across one or more channels includes extracting a first set of audio fingerprints and a first set of video fingerprints. The method also includes generating a set of digital signature values corresponding to an extracted set of video fingerprints, and normalizing each frame of a pre-determined number of frames of a video. The method also includes scaling each frame of the corresponding pre-determined number of frames of the video clip to a pre-defined scale. Each frame corresponds to the broadcasted media content on the channel. The method also includes trimming a first pre-defined region and a second pre-defined region of each frame by a pre-defined percentage of a frame width, a frame height and a pre-defined number of pixels in each frame.