Adaptive Wavelet Feature Pool for Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object detection methods using learning-based classifiers are limited in their ability to adaptively select features from training image samples, leading to suboptimal performance in distinguishing objects from non-objects.
Innovation Solution
The method involves initializing and optimizing wavelet parameters to match image structures in training samples, generating an adaptive feature pool, and using a classifier based on common characteristics to detect objects in images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed set of features is used for object detection, then the classifier can be trained efficiently, but the system cannot adapt to different image structures and object variations
Solution Approach 1:
The patent applies dynamics by making the feature pool adaptive rather than fixed. Wavelet parameters (scale, orientation, position) are dynamically adjusted to match image structures during training. The system transitions from a static feature set to a dynamic one that evolves based on training data characteristics, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The patent changes wavelet parameters (scale, orientation, position) based on image structure analysis. By modifying these parameters during the training process, the system creates an adaptive feature pool that tailors features to specific image characteristics, thereby improving adaptability while managing complexity through parameter optimization.
2Measurement precision
If adaptive wavelet parameters are optimized for each training sample, then detection accuracy improves, but computational complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-optimizing wavelet parameters during the training phase. The computationally intensive parameter optimization is done beforehand to create an adaptive feature pool, so that during actual detection, the system can use these pre-computed features efficiently without repeating the full optimization process, thus balancing accuracy with computational power.
Solution Approach 2:
The patent applies partial action by optimizing wavelet parameters selectively for each training sample rather than exhaustively for all possible parameters and samples. This targeted optimization approach achieves sufficient detection accuracy while reducing the overall computational burden compared to a complete exhaustive search.
3Reliability
If a large pool of adaptive features is created, then the classifier can distinguish objects better, but the training time and processing overhead increase
Solution Approach 1:
The patent extracts only the most relevant wavelet parameters and features that are necessary for accurate object detection. Rather than using all possible wavelet configurations, the system selects and extracts the essential adaptive features that provide the best discrimination between objects and non-objects, thereby improving reliability while reducing training time and processing overhead.
Data Source
AI summary
A method, system and computer program product for detecting presence of an object in an image are disclosed. According to an embodiment, a method for detecting a presence of an object in an image comprises: receiving multiple training image samples; determining a set of adaptive features for each training image sample, the set of adaptive features matching the local structure of each training image sample; integrating the sets of adaptive features of the multiple training image samples to generate an adaptive feature pool; determining a general feature based on the adaptive feature pool; and examining the image using a classifier determined based on the general feature to detect the presence of the object.