Motion Recognition Using Depth and 2D Data Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for recognizing motion of objects using two-dimensional and depth data face challenges in distinguishing objects from backgrounds due to lower resolution depth data and difficulty in discerning complex shapes, leading to inefficient motion recognition.
Innovation Solution
A method and apparatus that periodically obtain depth data and two-dimensional data at different resolutions, synchronize them, and extract tracking region data from a motion tracking region to analyze the motion of objects efficiently, enhancing resolution through image data compensation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth data is used for motion recognition, then object distance information is obtained, but resolution is low making it difficult to discern complex shapes
Solution Approach 1:
The patent combines depth data and two-dimensional image data into a unified motion recognition process. The depth information provides distance measurement while the two-dimensional data provides shape details, and their integration allows the system to achieve both depth measurement precision and shape discernment accuracy that neither data type could provide alone.
2Manufacturing precision
If two-dimensional data is used for motion recognition, then shape and color information is obtained, but object separation from background is difficult
Solution Approach 1:
The patent applies motion recognition to a specific region of interest rather than analyzing the entire image frame. By focusing computational resources on a partially defined region where the object is located, the system achieves effective object-background separation without the complexity of processing all pixels in the two-dimensional data.
3Manufacturing precision
If high resolution data is processed for motion recognition, then detailed object information is obtained, but data transfer and calculation requirements increase
Solution Approach 1:
The patent segments the data processing into distinct stages: depth data is processed first to identify object location and define a region of interest, then only the two-dimensional data within that specific region is processed for detailed motion analysis. This segmentation allows high-resolution processing only where necessary, reducing overall data transfer and calculation requirements while maintaining detailed object information where needed.
Data Source
AI summary
A method of recognizing motion of an object may include periodically obtaining depth data of a first resolution and two-dimensional data of a second resolution with respect to a scene using an image capturing device, wherein the second resolution is higher than the first resolution; determining a motion tracking region by recognizing a target object in the scene based on the depth data, such that the motion tracking region corresponds to a portion of a frame and the portion includes the target object; periodically obtaining tracking region data of the second resolution corresponding to the motion tracking region; and/or analyzing the motion of the target object based on the tracking region data.