CNN Feature Extraction for Secure Image Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition technologies face issues with data security as image data transmitted from cameras to servers can be leaked during the transmission process.
Innovation Solution
Implementing a recognition system where an image capturing device generates and transmits feature information instead of raw image data, using a convolutional neural network (CNN) to process the data and an iterative method to generate recognition results, thereby preventing data leakage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image data is transmitted from camera to server for recognition, then recognition accuracy is improved, but data security deteriorates due to potential leakage during transmission
Solution Approach 1:
The patent extracts only the essential feature information from the complete image data using CNN processing. Instead of transmitting the entire image, only the extracted features (such as edge information, texture features, or high-level semantic features) are sent to the server. This extraction principle maintains recognition accuracy while minimizing data transmission volume and leakage risk.
Solution Approach 2:
The patent performs preliminary processing of image data at the camera端 before transmission. The CNN model processes the image data locally to generate feature information in advance, so that when data is transmitted to the server, it is already in a processed, compressed feature format rather than raw image data. This preliminary action reduces both transmission bandwidth requirements and security risks.
2Object-affected harmful factors
If feature information is transmitted instead of image data, then data security is improved, but processing complexity increases due to iterative calculation requirements
Solution Approach 1:
The patent segments the CNN processing tasks between two locations: the camera端 performs early layer processing (feature extraction from raw images), while the server端 performs later layer processing (iterative calculation to generate final recognition results). This segmentation distributes computational complexity across different devices, reducing the burden on any single device while maintaining overall system security and functionality.
Solution Approach 2:
The patent introduces feature information as an intermediary between the camera端 and server端. Instead of directly transmitting raw image data or requiring the server to perform all processing, the feature information serves as a middle representation that preserves essential information while reducing security risks and transmission requirements. The server then uses this intermediary feature data to complete the recognition process through iterative calculation.
3Productivity
If iterative method is used to generate recognition results, then processing efficiency is improved, but computational resources increase
Solution Approach 1:
The patent divides the CNN computational workload into segments executed at different locations and times. The camera端 performs early layer computations locally using available resources, then transmits intermediate feature results to the server. The server performs the remaining iterative calculations using its computational resources. This segmentation allows the system to leverage distributed computational resources efficiently, improving overall processing efficiency while managing energy consumption across the system.
Data Source
AI summary
A recognition system includes an image capturing device and a server. The image capturing device generates a Mth-layer calculation result based on image data and a convolutional neural network (CNN), and transmits feature information associated with the Mth-layer calculation result. M is a positive integer, M is equal to or greater than 1, M is less than or equal to N, and N is a predetermined positive integer. The server receives the feature information. The server generates a Kth-layer calculation result based on the feature information and the CNN by an iterative method when M is less than N. K is a positive integer which is greater than M and less than or equal to N. The server generates a first recognition result associated with the image data based on the Kth-layer calculation result and a first recognition model when K is equal to N.