High-Canopy Wetland Forest Parameter Extraction Using Consumer UAVs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current forest resource investigation methods using UAV visible light images face challenges such as high costs and operation requirements for professional mapping UAVs, low positioning accuracy of consumer-grade UAVs, single-band visible light images limiting vegetation index calculations, and lower accuracy in segmenting high-canopy density wetland vegetation.
Innovation Solution
A method utilizing a consumer-grade UAV like the DJI Mavic 2 Zoom to collect image data, perform georeferencing and aerial triangulation, and generate DSM and DOM models. This involves land use classification, extraction of elevation points, regeneration of DTM, and obtaining CHM to accurately determine tree height and number. Additionally, a multi-foreground marker watershed algorithm based on image bilateral filtering is used for accurate crown segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a professional mapping UAV with RTK module is used, then positioning accuracy and spatial coordinate precision are improved, but device cost and operation cost increase significantly
Solution Approach 1:
The patent replaces expensive professional mapping UAVs with RTK modules with affordable consumer-grade UAVs without RTK. The solution accepts lower positioning accuracy from consumer-grade UAVs but compensates through post-processing techniques (aerial triangulation optimization, water surface reflection removal) to achieve acceptable measurement precision at significantly reduced device cost
Solution Approach 2:
The patent changes the operational parameters of consumer-grade UAVs by optimizing flight height, image overlap rates, and processing parameters for aerial triangulation. It also applies parameter optimization in the image processing stage (vegetation index thresholds, segmentation parameters) to compensate for the lack of RTK positioning accuracy
2Ease of manufacture
If consumer-grade UAV without RTK module is used, then device cost is reduced, but positioning accuracy and spatial coordinate precision deteriorate
Solution Approach 1:
The patent introduces an intermediary processing system that acts as a mediator between the consumer-grade UAV's low-precision positioning and the final measurement requirements. The aerial triangulation optimization, water surface reflection removal, and vegetation index calculation processes serve as intermediary steps that improve the effective positioning accuracy without requiring expensive RTK hardware
Solution Approach 2:
The patent creates a corrected digital surface model (DSM) and digital orthophoto map (DOM) as accurate copies that compensate for the original low-precision positioning. Through iterative optimization and error correction in processing, it generates high-quality output products that meet measurement requirements despite using lower-cost input data from consumer-grade UAVs
3Ease of manufacture
If visible light sensor with single band is used, then device cost is reduced, but vegetation index calculation capability is limited
Solution Approach 1:
The patent makes the visible light sensor perform multiple functions by calculating various vegetation indices (NDVI, EVI, SAVI, MSAVI) from the single-band RGB data through mathematical transformations. The system achieves multi-functionality in vegetation analysis despite the sensor's single-band limitation, enabling diverse ecological studies with a cost-effective device
Solution Approach 2:
The patent transforms the single-band visible light data into multiple vegetation indices through parameter changes and mathematical calculations. By deriving NDVI, EVI, SAVI, and MSAVI from RGB bands using different formulaic approaches, it expands the analytical capability of the simple sensor to match some functions of more expensive multi-spectral sensors
4Device complexity
If traditional marker watershed algorithm is used for crown segmentation, then processing simplicity is maintained, but segmentation accuracy in high canopy density areas deteriorates
Solution Approach 1:
The patent applies preliminary actions before crown segmentation by first calculating multiple vegetation indices, performing land use classification, and generating optimized input data. These preliminary processing steps prepare the data in advance to improve the effectiveness of subsequent segmentation algorithms, addressing the limitations of traditional methods in high-density canopy areas
Solution Approach 2:
The patent combines multiple processing techniques and data sources (RGB images, multiple vegetation indices, land use classification results, and improved segmentation algorithms) to create a composite processing system. This composite approach integrates the strengths of different methods to achieve high segmentation accuracy in complex high-density canopy conditions while managing processing complexity
Data Source
AI summary
A method for extracting forest parameters of a wetland with high canopy density based on a consumer-grade unmanned aerial vehicle (UAV) image is provided, which belongs to the technical field of wetland forest investigation. The method includes: collecting image data using a consumer-grade UAV; processing the collected UAV image, and generating a digital surface model and a digital orthophoto map; performing land use classification based on a UAV DOM image; obtaining a canopy height model with an accurate elevation based on classification of different land types of the wetland, and obtaining an accurate tree height and a tree number by means of a neighborhood extraction tool through the canopy height model; and obtaining an accurate canopy range by means of a multi-foreground marker watershed algorithm based on image filtering.