Trace Transform Image Identification via Multi-Resolution Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image identification techniques face challenges in achieving high detection rates and robustness to noise and histogram equalization modifications, with detection rates below 98% and limited resistance to image alterations.
Innovation Solution
The method involves deriving a reduced resolution Trace transform of an image using selected lines or intervals, followed by a multi-resolution representation through region-based processing and Fourier Transform, to create a binary descriptor that is robust to various image modifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional image identification techniques are used, then computational complexity is reduced, but detection rate remains below 98% and robustness to noise and histogram equalization is limited
Solution Approach 1:
The patent divides the image processing into multiple resolution levels. The Trace transform is computed at different resolutions, and binary descriptors are extracted at each level. This multi-resolution segmentation allows the system to capture both coarse and fine details, improving detection rate while managing computational complexity through hierarchical processing.
Solution Approach 2:
The patent introduces a multi-resolution dimension by computing the Trace transform at multiple scales. Instead of processing the image at a single resolution, the system processes it at several resolution levels, adding a dimensional aspect to the feature extraction that improves robustness and detection rate without linearly increasing computational burden.
2Reliability
If traditional binary descriptors are used, then computational complexity is low, but robustness to histogram equalization and noise modifications is insufficient
Solution Approach 1:
The patent applies preprocessing steps including Trace transform and multi-resolution decomposition before extracting binary descriptors. This preliminary action of transforming the image into a different domain (Trace transform) and processing at multiple resolutions beforehand makes the subsequent descriptor extraction more robust to modifications like histogram equalization and noise, as these operations affect the original image domain differently than the transform domain.
3Measurement precision
If full resolution Trace transform is computed, then detection accuracy is maximized, but computational complexity and processing time increase
Solution Approach 1:
The patent segments the Trace transform computation into multiple resolution levels. Instead of computing the full-resolution Trace transform, the system computes it at several lower resolution levels, extracting binary descriptors at each level. This segmentation maintains detection accuracy by capturing features at different scales while significantly reducing the computational burden and processing time.
Solution Approach 2:
The patent uses partial action by computing the Trace transform only at selected resolution levels rather than at all possible resolutions. This selective computation provides sufficient detection accuracy for practical applications while avoiding the excessive computational cost of full-resolution processing, achieving a practical balance between accuracy and efficiency.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly improves detection rates to above 98% and enhances robustness to noise and histogram equalization, while reducing computational complexity and false alarm rates, achieving a low false-alarm rate of 1 part per million.
Implementation Method 1
a multi-resolution representation of an image by performing region-based processing on the Trace transform of the image, prior to extraction of the identifier e.g. by means of the magnitude of the Fourier Transform
Data Source
Figure 1(a)~1(d)
Figure 2
Figure 3
AI summary
A method and apparatus for deriving a representation of an image by processing signals corresponding to the image is described. The method includes deriving a two-dimensional function (T(d, ?)), such as a Trace transform of the image, and decomposing, for instance by sub-sampling, the two-dimensional function (T(d, ?)) in at least one of its two dimensions, to obtain a reduced resolution Trace transform. The decomposed, two dimensional function is then used to derive the representation of the image.