3D Map Creation Excluding Moving Object Voxels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current SLAM technologies fail to accurately form three-dimensional maps when input images contain moving objects, as they assume stationary objects, leading to incorrect map formation and the need for map recreation upon object movement.
Innovation Solution
An image processing device and method that restricts three-dimensional position recognition of pixels based on attached labels, dividing the image into regions for separate recognition and using weight coefficients to minimize errors, allowing for accurate map creation even with moving objects by excluding their voxels from the map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional position recognition is executed for all pixels in the captured image, then the three-dimensional map can be formed, but the map becomes inaccurate when moving objects are included
Solution Approach 1:
The captured image is divided into multiple regions based on object type recognition results. Pixels are classified into different regions (e.g., stationary objects, moving objects, background) and three-dimensional position recognition is selectively applied to specific regions, excluding moving objects from map formation.
Solution Approach 2:
Different processing strategies are applied to different regions of the image. Stationary object regions undergo three-dimensional position recognition and contribute to map formation, while moving object regions are excluded or handled differently, ensuring local optimization of map accuracy.
2Measurement precision
If the three-dimensional map is recreated whenever an object is moved, then the map remains accurate, but the processing time and computational load increase
Solution Approach 1:
Object type recognition is performed in advance on the captured image to identify moving objects before three-dimensional position recognition. This preliminary classification allows the system to selectively exclude moving objects from map formation, avoiding the need for complete map recreation when objects move.
Solution Approach 2:
The system dynamically adjusts the three-dimensional position recognition process based on real-time object detection results. When moving objects are detected, the system adaptively modifies the processing to exclude these regions, maintaining map accuracy without requiring full map recreation.
3Reliability
If object type recognition is performed for each pixel, then moving objects can be identified, but the computational complexity increases
Solution Approach 1:
The object type recognition system serves multiple functions: it identifies moving objects, classifies pixels into different regions, and provides guidance for selective three-dimensional position recognition. This multi-functional approach reduces the need for separate processing systems and minimizes overall computational complexity.
Data Source
AI summary
Provided are an image processing device, an image processing method, and a program capable of forming an accurate three-dimensional map even in a case where a moving object is included in an input image. The image processing device includes an image acquiring section that sequentially acquires two-dimensional captured images, an object type recognition executing section that attaches, to each pixel in the sequentially acquired captured images, a label indicating the type of an object represented by the pixel, and a three-dimensional map creating section that executes three-dimensional position recognition of each pixel of the captured images to create a three-dimensional map, on the basis of the sequentially acquired captured images, and the three-dimensional map creating section restricts the three-dimensional position recognition of each pixel of the captured images according to the label attached to the pixel.