Laser Scanner Image Correction via Camera-Lidar Discrepancy Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Laser scanning devices face challenges in generating accurate 3D images due to errors caused by objects moving through the scene during data capture, leading to erroneous data and noise in the final point cloud and image sets.
Innovation Solution
The method involves using a laser scanner with cameras to detect moving objects by comparing pixel differences in camera images and intensity data from laser scans, identifying decision regions, and generating corrected images by selecting the camera image that best matches the laser scan data, while removing moving objects through image processing and blending techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If laser scanning is performed to capture 3D data, then comprehensive scene data is obtained, but moving objects cause errors and noise in the point cloud and image sets
Solution Approach 1:
The system performs preliminary comparison between multiple camera images and laser scan data before final image generation. By identifying decision regions and comparing camera images against laser scan data in advance, the system detects moving objects and selects the most accurate data sources before constructing the final point cloud and image sets, thereby preventing errors from propagating into the final output
Solution Approach 2:
The system uses feedback mechanisms by comparing camera images with laser scan data and using the detected edges and decision regions to iteratively refine the selection of accurate data. The comparison results feed back into the image generation process, allowing the system to continuously improve accuracy by selecting data from time points with minimal moving object interference
2Measurement precision
If multiple camera images are captured to identify moving objects, then accuracy is improved, but processing complexity increases
Solution Approach 1:
The system segments the image processing task by first comparing camera images to identify specific decision regions where differences occur, then focusing edge detection and moving object analysis only on those regions. This segmentation approach divides the complex task of analyzing entire images into manageable regional comparisons, reducing overall processing complexity while maintaining detection accuracy
Solution Approach 2:
The system applies local quality by treating different regions of images differently - decision regions are subjected to rigorous comparison and edge detection, while stable regions are accepted without additional processing. This localized approach concentrates computational resources where they are most needed (in decision regions with potential moving objects) and avoids unnecessary processing in stable areas
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively removes moving objects from laser scan data, resulting in cleaner, more accurate 3D images by automatically identifying and correcting discrepancies between camera images and laser scan data, enhancing the quality of the final point cloud and image sets.
Implementation Method 1
a laser scanner with a laser rangefinder configured to acquire a three-dimensional point cloud of a scene
Data Source
AI summary
In one example, a method may include receiving one or more laser scan of a scene, receiving two or more camera images of the scene, determining one or more decision regions where the camera images are different from one another, detecting edges of the decision regions where the camera images are different from one another, comparing the decision regions where the camera images are different from one another inside of the detected edges with a corresponding region in the laser scan to determine which of the camera images includes a desired region that more closely corresponds to the laser scan, and generating a corrected image including the desired region that more closely corresponds to the laser scan.