Image Processing Device for Event Camera Data Transfer Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Event cameras generate high-speed data output but face increased data transfer amounts due to object motion, particularly when transferring data to main memory like DRAM.
Innovation Solution
An image processing device is used in conjunction with an event camera to acquire pixel data, including luminance changes and two-dimensional address values, which are converted into one-dimensional address values and used to calculate scanning distances. This information is then used to generate binary data with varying bit lengths, reducing data transfer by discarding or recalculating scanning distances and using escape codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If event camera outputs data for all pixels with luminance changes, then motion detection speed is improved, but data transfer amount increases
Solution Approach 1:
The patent extracts only the essential information needed for motion detection by calculating scanning distances between consecutive luminance change events. Instead of transferring all raw pixel data, only the scanning distance values are transferred to the main memory, filtering out redundant data while preserving motion detection capability
Solution Approach 2:
The patent transforms two-dimensional pixel coordinate data into one-dimensional scanning distance values by mapping pixel positions along the scanning direction. This dimensional reduction converts (x, y) coordinate pairs into single distance values, significantly reducing data transfer requirements while maintaining motion information
Data Source
AI summary
An image processing device is used in combination with a camera configured to acquire pixel data including a two-dimensional address value indicating a pixel position and a luminance change amount for each pixel of which a change in luminance is detected by scanning a plurality of pixels in a predetermined scanning order, and includes an address value acquisition unit, a scanning distance calculation unit, and a binary data generation unit. The address value acquisition unit converts the two-dimensional address value to acquire a one-dimensional address value. The scanning distance calculation unit calculates a scanning distance from a first pixel to a second pixel based on the one-dimensional address value. The binary data generation unit generates binary data including bit strings that have different number of bits according to a calculation result of the scanning distance and a bit string indicating the luminance change amount.