AI Acoustic Camera for Abnormal Source Localization and Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing acoustic cameras struggle to accurately determine the position and type of acoustic sources, particularly abnormal sources, and fail to integrate image analysis for object recognition, limiting their effectiveness in identifying specific acoustic events.
Innovation Solution
An AI acoustic image camera that utilizes an acoustic sensor array to localize acoustic sources, extract time-domain acoustic signals, generate feature images, and perform AI classification to identify abnormal sources and associated objects, integrating video analysis for precise determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If acoustic camera systems use MEMS microphone arrays with analog signal processing, then the device structure is relatively simple, but the accuracy in determining position and type of abnormal acoustic sources is insufficient
Solution Approach 1:
The patent replaces traditional analog signal processing mechanisms with AI-based image analysis mechanisms. Acoustic signals are transformed into visual feature images (spectrograms, heatmaps) that can be processed by computer vision algorithms, enabling the system to not only locate but also classify abnormal acoustic sources with high accuracy using pre-trained neural network models.
Solution Approach 2:
The patent transforms acoustic signal data from the time-frequency domain into visual image representations (spectrograms, heatmaps, feature maps). This dimensional transformation allows the application of powerful 2D image processing and classification techniques to acoustic data, significantly improving the ability to identify and classify abnormal sources beyond simple location detection.
2Adaptability or versatility
If acoustic camera systems perform basic acoustic source localization, then the processing speed is relatively fast, but the ability to recognize and classify the type of acoustic source is lacking
Solution Approach 1:
The patent employs pre-trained AI models (such as CNNs for image classification) that have been previously trained on large datasets of acoustic feature images. This preliminary training allows the system to perform rapid classification during actual operation without requiring real-time complex computations, thus maintaining high processing speed while achieving advanced source type recognition capabilities.
Solution Approach 2:
The patent uses visual feature images (spectrograms, heatmaps) as copies or representations of the original acoustic signals. These image copies can be processed by efficient 2D convolutional neural networks that leverage spatial relationships in the visual domain, enabling fast classification while capturing complex temporal and spectral characteristics of different acoustic sources.
3Measurement precision
If acoustic camera systems use advanced AI image analysis for object recognition, then the accuracy of source identification improves, but the device complexity and computational requirements increase
Solution Approach 1:
The patent extracts only the most relevant features from acoustic signals by transforming them into focused visual representations (spectrograms with specific frequency ranges, heatmaps highlighting abnormal regions). This extraction of essential features reduces the dimensionality and complexity of data that needs to be processed by AI models, maintaining high identification accuracy while reducing computational burden.
Solution Approach 2:
The patent employs universal AI models (such as pre-trained CNN architectures) that can be applied to various types of acoustic source classification tasks. These multi-functional models can identify different types of abnormal sources (gas leaks, mechanical failures, electrical anomalies) using the same underlying technology framework, reducing overall system complexity compared to having specialized models for each source type.
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
Enables accurate localization and identification of abnormal acoustic sources and associated objects, enhancing the camera's ability to detect and alert for specific acoustic events like gas leakage by correlating acoustic and visual data.
Implementation Method 1
acoustic data acquired by a plurality of acoustic sensor array
Data Source
AI summary
Disclosed is an AI acoustic camera including an acoustic source localizing means unit of generating position-specific acoustic level data by determining a position of an acoustic source, an AI acoustic analysis unit of recognizing a type of acoustic source estimated as an abnormal acoustic source by extracting a regeneration time domain acoustic signal for the acoustic source with the determined position and AI-learning and recognizing an acoustic feature image of the extracted time domain acoustic signal, an object recognition unit of recognizing a type of object positioned in the acoustic source through image analysis of an area recognized as that the acoustic source is positioned, and a determination unit of determining the acoustic source as a true acoustic source when the type of acoustic source and the type of object have commonality.