LiDAR Point Cloud Encoding Using Angle Model Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing LiDAR point cloud compression methods, such as G-PCC and Japanese Unexamined Patent Application 2018-116452, face efficiency issues when dealing with complex scanning paths and require templates of geographical features, leading to decreased compression efficiency.
Innovation Solution
An encoding apparatus that calculates predicted points based on measurement model information and encodes differences between predicted and measured points, using an angular coding mode to reduce information amount and maintain compression efficiency even with complex scanning paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If G-PCC encoding method is used for point cloud compression, then compression capability is provided, but compression efficiency decreases when LiDAR has complicated scanning paths
Solution Approach 1:
The patent applies preliminary action by pre-storing the scanning path information and using it to generate prediction values for azimuth and elevation angles before actual encoding. This allows the encoder to predict what the angles should be based on the known scanning pattern, and then encode only the differences from these predictions, significantly improving compression efficiency for complicated scanning paths.
Solution Approach 2:
The patent implements feedback by calculating the difference between actual measured angles and predicted angles from the scanning path, then using this difference information for encoding. The decoding side reverses this process by adding the predicted values and difference values to reconstruct the original angles, ensuring accurate reconstruction while maintaining high compression efficiency.
2Quantity of substance
If template-based compression method is used, then compression is achieved, but the method is easily influenced by surrounding environment
Solution Approach 1:
The patent extracts the scanning path information from the point cloud data and uses it independently to generate prediction values. This separates the compression process from environmental influences, as the prediction is based solely on the known scanning pattern rather than templates that may not match the actual scanning conditions. This extraction approach ensures reliable compression regardless of surrounding environment.
3Loss of information
If point cloud data with multiple attributes is processed, then comprehensive information is captured, but storage and communication costs increase
Solution Approach 1:
The patent extracts and processes only the angular information (azimuth and elevation angles) from the point cloud data for compression, while other attributes like time stamp, intensity, and RGB information are handled separately or excluded from this particular compression process. This selective extraction maintains the necessary spatial information for 3D reconstruction while significantly reducing the data volume that requires compression.
Data Source
AI summary
In order to improve encoding efficiency in cases in which a scan is performed by irradiating a laser, or the like, an encoding apparatus includes an acquisition unit configured to acquire point cloud data consisting of a plurality of measured points from a measurement unit; a calculating unit configured to calculate predicted points corresponding to the plurality of measured points based on measurement model information relating to a measurement pattern of the measurement unit; a difference calculating unit configured to calculate a difference between the measured points and the predicted points corresponding to the measured points; and an encoding unit configured to encode the difference.