Dominant Color Detection Using Trained Classification Detectors

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current techniques for identifying the dominant color of an image often fail to accurately select the appropriate color, especially when an image has a significant portion of its pixels in one color and smaller amounts in very similar colors, leading to incorrect dominant color identification.

Innovation Solution

A method and system that trains a detector using a collection of training images to classify colors as dominant by extracting features based on differences within the image and applying classification techniques like support vector machines, adaptive boosting, and neural networks to identify the dominant color of target images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current techniques select the color that appears most frequently in an image using statistical algorithms, then the identification process is simple and fast, but the accuracy of dominant color identification deteriorates when the image has a significant portion of pixels in one color and smaller amounts in very similar colors

Engineering Contradiction:

Improvedominant color identification accuracyVSAvoidcolor classification system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary actions by collecting training images and pre-training the detector with labeled dominant color information before actual dominant color identification is needed. This pre-processing stage prepares the classification model to handle various color distribution patterns, improving accuracy without adding complexity to the real-time identification process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces simple statistical algorithms with a machine learning-based detector that uses classification techniques. Instead of relying on basic pixel counting and frequency analysis, the system employs trained classifiers that can distinguish between dominant colors and very similar colors by learning from training data, thereby improving measurement precision.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If a detector is trained using a collection of training images with classification techniques, then the dominant color identification accuracy is improved, but the system complexity and training time increase

Engineering Contradiction:

Improvedominant color identification reliabilityVSAvoiddetector training time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs detector training in advance as a preliminary action before actual dominant color identification tasks. By pre-training the detector with a collection of training images and their corresponding dominant color labels, the system establishes a reliable classification model that can be reused for multiple images without repeating the time-consuming training process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent employs parameter changes by adjusting classification thresholds and using multiple classification techniques (such as support vector machines, neural networks, or random forests) to optimize the detector's performance. These parameter adjustments improve reliability by making the classifier more sensitive to subtle color differences while maintaining efficiency.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If statistical algorithms are used to identify dominant colors, then the processing is computationally efficient, but the system fails to accurately distinguish between a dominant color and very similar colors that appear in smaller amounts

Engineering Contradiction:

Improvecolor distinction accuracyVSAvoidcomputational energy consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent substitutes simple statistical algorithms with a machine learning detector that uses classification techniques. This replacement enables the system to accurately distinguish between dominant colors and very similar colors by learning from training data, achieving high measurement precision through intelligent classification rather than basic computational methods.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system uses parameter changes by adjusting classification thresholds and employing multiple classification techniques to optimize the balance between accuracy and computational efficiency. The trained detector learns optimal parameter settings during training, enabling accurate color distinction while maintaining reasonable computational energy consumption during actual operation.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS7809185B2Extracting dominant colors from images using classification techniques

Publication Date: 2010.10.05 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

A method and system for generating a detector to detect a dominant color of an image is provided. A dominant color system trains a detector to classify colors as being dominant colors of images. The dominant color system trains the detector using a collection of training images. To train the detector, the dominant color system first identifies candidate dominant colors of the training images. The dominant color system then extracts features of the candidate dominant colors. The dominant color system also inputs an indication of dominance of each of the candidate dominant colors. The dominant color system then trains a detector to detect the dominant color of images using the extracted features and indications of dominance of the candidate dominant colors as training data.