Compressed-Domain Camera Attribute Setting Without Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Compressive sensing in digital imaging reduces battery usage and storage needs but requires computationally intensive reconstruction to retrieve the original image, making it challenging for real-time implementation of automatic camera features like autofocus in devices.
Innovation Solution
A compressive sensing capturing device and method that uses a machine learning algorithm, specifically a convolutional neural network, to set device attributes directly from compressed image data, eliminating the need for image reconstruction and enabling faster, battery-efficient automatic features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If compressive sensing is used to reduce measurements and storage, then battery usage and storage requirements are reduced, but computationally intensive reconstruction is required to retrieve the original image
Solution Approach 1:
The patent pre-calculates and stores measurement matrices during device manufacturing or initialization. These pre-computed matrices are stored in memory and reused during operation, eliminating the need for real-time computation of measurement matrices and reducing processing power requirements during actual image capture and reconstruction
Solution Approach 2:
The patent uses multiple measurement matrices that can be randomly selected from a stored set, allowing the system to reuse proven effective measurement patterns without重新 computing them. This copying approach reduces computational overhead while maintaining reconstruction quality
2Quantity of substance
If compressive sensing is used to reduce measurements, then storage requirements are reduced, but computationally intensive reconstruction is required to retrieve the original image
Solution Approach 1:
The patent pre-computes and stores multiple measurement matrices in memory during device initialization. These pre-stored matrices are readily available for selection during operation, eliminating the need for complex real-time computation and reducing both storage and processing requirements during actual use
Solution Approach 2:
The patent changes the sparsity domain parameters by allowing selection from multiple pre-computed measurement matrices corresponding to different sparsity domains. This enables adaptation to different image characteristics without requiring intensive real-time computation of new measurement matrices
3Measurement precision
If computationally intensive reconstruction is performed to retrieve the original image, then image quality is improved, but real-time implementation of automatic camera features becomes challenging
Solution Approach 1:
The patent pre-calculates measurement matrices and stores them in memory during device initialization or manufacturing. This preliminary computation eliminates the need for intensive real-time matrix computation during image capture and reconstruction, enabling faster processing while maintaining image quality
Solution Approach 2:
The patent extracts and separates the computationally intensive matrix computation step from the real-time operation. By moving the computation to a preliminary stage and storing results, the system removes the computational burden from real-time processing, enabling both high image quality and real-time performance
Data Source
AI summary
A compressive sensing capturing device has circuitry, which is configured to obtain compressive sensing image data; and to set a device attribute based on image attribute data, wherein the image attribute data are based on a machine learning algorithm performed in the compressing domain on the obtained compressive sensing image data.