Dynamic Vision Sensor Object Recognition via Frame Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Devices for recognizing objects using RGB images or dynamic vision sensors face challenges in accuracy due to external lighting changes and privacy concerns, with incorrect bit data collection and background interference.
Innovation Solution
An electronic device equipped with a dynamic vision sensor, processor, and memory filters out partial image frames based on bit data ratios to enhance object recognition accuracy, using illuminance and motion sensors to exclude frames affected by lighting changes and background noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If DVS is used to generate shape images for object recognition, then privacy is protected, but recognition accuracy deteriorates due to incorrect bit data collection from lighting changes and background objects
Solution Approach 1:
The system performs preliminary actions by collecting both RGB image data and bit data before recognition, pre-processing the data to identify and remove frames with lighting changes or device movement, ensuring only valid frames are used for subsequent shape-based recognition
Solution Approach 2:
The system introduces an intermediary validation mechanism that uses RGB images as a reference to verify bit data quality. The RGB image serves as a mediator to detect lighting changes and device movement, allowing the system to filter out unreliable bit data while maintaining privacy protection through shape-based recognition
2Reliability
If all image frames are processed for object recognition, then recognition coverage is improved, but computational processing and power consumption increase
Solution Approach 1:
The system extracts and removes invalid frames from the processing stream by identifying frames with lighting changes or device movement through RGB image analysis, then excludes these frames from bit data processing and object recognition, reducing computational load while maintaining recognition coverage for valid frames
Solution Approach 2:
The system changes the processing parameter by dynamically adjusting the number of frames processed based on their validity. Valid frames are processed for object recognition while invalid frames are skipped, optimizing the balance between recognition coverage and power consumption
Data Source
AI summary
According to an embodiment of the disclosure, an electronic device may include a communication interface, a dynamic vision sensor (DVS) to generate bit data for each of a plurality of image frames, based on change in illuminance, a processor electrically connected with the communication interface and the DVS, and a memory electrically connected with the processor. The memory may store instructions, and the instructions may cause the processor to filter out at least a partial frame of the plurality of image frames, based on a ratio of the number of a bit value, which is included in the bit data and corresponds to each pixel, to the number of total pixels constituting each of the plurality of image frames, and recognize a shape of a surrounding object of the electronic device, based on another frame of the plurality of image frames. Moreover, various embodiment found through the disclosure are possible.