Trace Transform Image Identification Using Mask Function

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current image identification techniques face challenges in achieving a false alarm rate significantly lower than 1 ppm and maintaining detection rates above 98%, especially in the presence of image modifications such as histogram equalization and cropping.

Innovation Solution

The method involves extracting visual identification features from the Trace transform of an image, spatially restricting the Trace transform to derive additional binary component identifiers, which are combined with existing descriptors to enhance robustness and performance, and using a mask function to compute the circus function from bands in the Trace domain, thereby improving detection rates to 99.80% at a reduced false alarm rate of 0.1 ppm.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional image identification methods are used, then computational complexity is reduced and descriptor size is reduced, but false alarm rate cannot be lowered below 1 ppm level

Engineering Contradiction:
Improvefalse alarm rateVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the Trace transform representation into multiple subsets of lines, extracting binary component identifiers from each subset. This segmentation allows the system to process and compare multiple restricted representations rather than one complete representation, improving reliability by providing multiple opportunities for correct identification while maintaining computational efficiency through localized processing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts binary component identifiers from subsets of lines rather than processing the entire Trace transform. This partial action approach reduces computational complexity per component while generating multiple identifiers that can be combined, ultimately achieving lower false alarm rates through the cumulative effect of multiple partial extractions.

Inventive Principle:
Principle #16Partial or excessive action

2Reliability

If traditional image identification methods are used, then detection rate is maintained above 98%, but robustness to histogram equalisation and image cropping is insufficient

Engineering Contradiction:
Improverobustness to image modificationsVSAvoidmethod complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies spatial restriction to create subsets of lines from the Trace transform, where each subset focuses on specific regions or orientations. This local quality approach ensures that modifications in one part of the image (such as cropping or histogram equalisation) do not completely destroy the identifier, as other restricted representations remain intact and can still provide reliable detection.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent combines multiple binary component identifiers extracted from different subsets of lines into a composite identifier. This composite approach enhances robustness because the combined identifier leverages information from multiple independent subsets, making it more resistant to various image modifications than any single subset alone.

Inventive Principle:
Principle #40Composite materials

3Reliability

If additional binary component identifiers are extracted from spatially restricted Trace transform, then detection rate increases to 99.80%, but false alarm rate reduction to 0.1 ppm requires more processing

Engineering Contradiction:
Improvedetection rateVSAvoidprocessing efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent segments the Trace transform into multiple subsets and extracts binary component identifiers from each segment in parallel. This segmentation enables efficient processing by distributing the computational workload across multiple independent extractions, achieving high detection rates through the combination of multiple identifiers without proportionally increasing overall processing time.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs spatial restriction and subset creation as preliminary actions before extracting binary component identifiers. By pre-organizing the Trace transform into meaningful subsets based on spatial or directional criteria, the system prepares the data structure to enable rapid extraction of multiple identifiers, improving processing efficiency while maintaining high detection rates.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2171644B1High performance image identification
Publication Date: 2013.08.07 MITSUBISHI ELECTRIC R&D CENTRE EUROPE BV
  • EP2171644B1 patent drawingFigure 1(a)~1(f)
  • EP2171644B1 patent drawingFigure 2
  • EP2171644B1 patent drawingFigure 3

AI summary

A method and apparatus for deriving a representation of an image is described. The method involves processing signals corresponding to the image. A two-dimensional function of the image, such as a Trace transform (T (d, θ)), of the image using at least one functional T, is derived and processed using a mask function (β) to derive an intermediate representation of the image, corresponding to a one-dimensional function. In one embodiment, the mask function defines pairs of image bands of the Trace transform in the Trace domain. The representation of the image may be derived by applying existing techniques to the derived one-dimensional function.