Wavelet Transform Object Detection in Low Resolution Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for automatic object recognition in digital images struggle with variations in object appearance, especially in low resolution images where contrast is not easily distinguishable, and require significant computational resources.
Innovation Solution
A method involving image separation, 2D Discrete Wavelet Transform for edge detection, and 1D Continuous Wavelet Transform to extract feature vectors that include scale and coefficient ranges, allowing for classification using a feature vector that can be compared to a database, even in real-time and multi-dimensional environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical modeling methods are used for object recognition, then recognition accuracy improves, but computational resource consumption increases
Solution Approach 1:
The patent segments the object recognition process into distinct phases: image acquisition, wavelet transform decomposition, feature extraction from specific coefficients, and classification. This segmentation allows the system to focus computational resources only on extracting discriminative features rather than processing entire images through complex statistical models, thereby reducing overall computational resource consumption while maintaining recognition accuracy.
Solution Approach 2:
The patent extracts specific wavelet coefficients (detail coefficients at different scales and positions) as features for object recognition, rather than using all image data. This extraction principle selects only the most informative features that capture object characteristics, eliminating redundant computational operations and reducing resource consumption while preserving recognition accuracy.
2Measurement precision
If complex statistical calculations are performed for object recognition, then recognition capability improves, but processing speed decreases
Solution Approach 1:
The patent performs wavelet transform decomposition as a preliminary action before feature extraction and classification. The wavelet transform pre-processes the image data into a structured format with coefficients organized by scale and position, making subsequent feature extraction more efficient. This preliminary structuring avoids repeated complex calculations during classification, thereby improving processing speed while maintaining recognition capability.
Solution Approach 2:
The patent replaces complex statistical modeling mechanisms with a wavelet-based feature extraction and classification system. The wavelet transform provides a mathematical framework that naturally captures multi-scale image features, substituting the need for iterative statistical calculations with a more direct computational approach that maintains recognition capability while improving processing speed.
3Measurement precision
If high resolution imaging is used for object detection, then detection accuracy improves, but radiation exposure increases in medical imaging
Solution Approach 1:
The patent changes the parameter of image resolution by applying wavelet transform to decompose the image into multiple scales. This allows the system to extract meaningful features from lower resolution images by analyzing patterns across different scales, thereby maintaining detection accuracy while enabling the use of lower radiation doses in medical imaging applications.
Solution Approach 2:
The patent adds the dimension of scale analysis through wavelet transform, converting a single-resolution detection problem into a multi-scale feature extraction problem. By analyzing image features across multiple scales rather than relying solely on high resolution, the system achieves detection accuracy equivalent to high resolution imaging while working with lower resolution (lower radiation) input images.
Data Source
AI summary
The invention is a method of using Wavelet Transformation and Artificial Neural Network (ANN) systems for automatic detecting and classifying objects. To train the system in object recognition different images, which usually contain desired objects alongside other objects are used. These objects may appear at different angles. Different characteristics regarding the objects are extracted from the images and stored in a data bank. The system then determines the extent to which each inserted characteristic will be useful in future recognition and determines its relative weight. After the initial insertion of data, the operator tests the system with a set of new images, some of which contain the class objects and some of which contain similar and/or dissimilar objects of different classification. The system learns from the images containing similar objects of different classes as well as from the images containing the class objects, since each specific class characteristic needs to be set apart from other class characteristic. The system may be tested and trained again and again until the operator is satisfied with the system's success rate of object recognition and classification.