Simultaneous Localization and Mapping Initialization via Rotation Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing simultaneous localization and mapping initialization methods are inefficient, leading to longer initialization times and lower accuracy due to insufficient parallax between continuous frame images, which affects applications in augmented reality, virtual reality, and automatic driving.
Innovation Solution
A method that preprocesses continuous frame images to remove the influence of rotation, screens initial key frames using an adaptive-sized sliding window, and performs simultaneous localization and mapping initialization based on these key frames to ensure sufficient parallax and improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If continuous frame images are used for initialization, then the initialization process is simple, but the parallax between frames is insufficient leading to low accuracy
Solution Approach 1:
The patent applies preliminary action by performing rotation removal preprocessing on continuous frame images before key frame screening. This preprocessing step eliminates rotational components from the images in advance, ensuring that when key frames are selected and matched, the parallax is sufficient for accurate initialization without requiring complex post-processing corrections
2Productivity
If traditional key frame screening is used, then the processing speed is fast, but the parallax between key frames is insufficient affecting accuracy
Solution Approach 1:
The patent performs rotation removal preprocessing before key frame screening to ensure that the selected key frames have sufficient parallax. By eliminating rotational components in advance, the screening process can efficiently select key frames based on translational motion criteria, guaranteeing both processing speed and adequate parallax for accurate initialization
Solution Approach 2:
The patent changes the parameter evaluation criteria for key frame screening by using rotation-removed images. Instead of screening based on original images that contain rotational components, the system screens based on preprocessed images where rotation has been eliminated, allowing for more accurate parallax assessment and better key frame selection
3Device complexity
If rotation influence is not removed, then the processing is simpler, but the reprojection error increases reducing accuracy
Solution Approach 1:
The patent extracts and removes the rotational component from continuous frame images through preprocessing. By separating and eliminating the rotation influence, the system retains only the translational motion information needed for accurate SfM initialization, thereby reducing reprojection error without significantly increasing overall processing complexity
Solution Approach 2:
The rotation removal is performed as a preliminary preprocessing step before subsequent SfM initialization operations. This advance elimination of rotational components prevents rotation-induced errors from propagating through the initialization pipeline, ensuring higher accuracy in reprojection without adding complexity to the core SfM algorithms
Data Source
AI summary
The application provides a method, a device, and a non-transitory computer-readable storage medium for simultaneous localization and mapping initialization. In the method, an operation of removing influence of rotation is performed to a predetermined number of continuous frame images that are obtained, and then initial key frames are screened, with a pre-built adaptive-sized sliding window, from the predetermined number of continuous frame images after the operation and the initial key frames are used for initialization.