Modified Cross-Correlation Template Matching

Solution Overview

Problem

Existing image processing systems face challenges in accurately and efficiently performing template matching due to the normalization cross-correlation function being too sensitive to intensity changes, leading to incorrect multiple matching regions, especially in uniform or similar structured images.

Innovation Solution

An improved similarity metric is introduced, modifying the normalized cross-correlation function with factors related to statistics of both the template and image, allowing for controlled discriminative ability, enabling accurate and efficient template matching by penalizing differences in mean and standard deviation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the normalized cross-correlation function is used for template matching, then the matching process is simple and fast, but it is too sensitive to intensity changes leading to incorrect multiple matching regions

Engineering Contradiction:

Improvetemplate matching speedVSAvoidmatching accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent modifies the normalized cross-correlation function by introducing a penalty term that changes the sensitivity parameter of the similarity metric. The modified function NCC(p, t) = (Σ(p_i - μp)(t_i - μt)) / (σp * σt * (1 + k * |μp - μt|)) includes a penalty factor k * |μp - μt| that dynamically adjusts based on the difference between mean intensities of the patch and template, reducing sensitivity to intensity changes while maintaining computational efficiency

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces an intermediary penalty term that mediates between the original cross-correlation measure and the final similarity score. This penalty term acts as a bridge that adjusts the matching result based on intensity statistics, preventing incorrect matches caused by intensity variations without requiring complete redesign of the matching algorithm

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If the normalized cross-correlation function is modified to include penalty terms for intensity differences, then the discriminative power and accuracy of template matching improve, but the computational complexity increases

Engineering Contradiction:

Improvematching accuracyVSAvoidsimilarity metric complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent modifies the normalized cross-correlation function by introducing a penalty term that changes the sensitivity parameter of the similarity metric. The modified function NCC(p, t) = (Σ(p_i - μp)(t_i - μt)) / (σp * σt * (1 + k * |μp - μt|)) includes a penalty factor k * |μp - μt| that dynamically adjusts based on the difference between mean intensities of the patch and template, reducing sensitivity to intensity changes while maintaining computational efficiency

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent applies a partial modification to the cross-correlation function by adding only the necessary penalty term for intensity differences rather than completely redesigning the similarity metric. This partial action approach adds computational complexity only where needed (in the penalty calculation) while preserving the efficient cross-correlation computation for the majority of the matching process

Inventive Principle:

Principle #16Partial or excessive action

3Measurement precision

If the normalized cross-correlation function is modified to include penalty terms for intensity differences, then the discriminative power and accuracy of template matching improve, but the processing time increases

Engineering Contradiction:

Improvematching accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent calculates the mean intensity (μp, μt) and standard deviation (σp, σt) of the patch and template beforehand as preliminary statistics. These pre-computed values are then reused in the penalty term calculation, avoiding redundant computations and minimizing the additional processing time required by the modified similarity metric

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent modifies the normalized cross-correlation function by introducing a penalty term that changes the sensitivity parameter of the similarity metric. The modified function NCC(p, t) = (Σ(p_i - μp)(t_i - μt)) / (σp * σt * (1 + k * |μp - μt|)) includes a penalty factor k * |μp - μt| that dynamically adjusts based on the difference between mean intensities of the patch and template, reducing sensitivity to intensity changes while maintaining computational efficiency

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10671881B2Image processing system with discriminative control

Publication Date: 2020.06.02 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

An image processing apparatus is described comprising a processor configured to access a template of image elements. The processor is configured to search an image which is larger than the template to find a region which is similar to the template, where similarity is measured using a similarity metric. The similarity metric comprises a normalized cross correlation function which is modified to include at least one factor related to a statistic of both the template and the region.