Aerial Image Stitching Using Multispectral Reference Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face difficulties in extracting feature values and stitching aerial images from areas with few buildings, such as farms and forests, which hampers the assessment of plant growth conditions through aerial photography.
Innovation Solution
An image processing system comprising an imaging apparatus, an illuminance sensor, and a calculation apparatus that uses multispectral sensors to capture images at various wavelengths, generating multispectral images, and employing a stitching process facilitated by reference images and inspection images to effectively extract feature values and stitch aerial images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing methods are used to extract feature values from aerial images of farms and forests, then the processing can be performed with standard equipment, but the feature value extraction becomes difficult and stitching fails due to lack of sufficient features
Solution Approach 1:
The patent segments the image processing task by generating multiple types of reference images (edge images, gradient images, Laplacian images) from the original aerial images. Each reference image type highlights different feature characteristics, making it easier to extract sufficient feature values even from areas with few buildings. This segmentation approach transforms the difficult feature extraction problem into multiple easier sub-tasks.
Solution Approach 2:
The patent transforms the 2D aerial images into multiple dimensional representations by creating different types of reference images with different mathematical transformations (edge detection, gradient calculation, Laplacian filtering). This dimensional transformation enriches the feature space, providing more extractable features from the same original image data.
2Reliability
If multiple reference images are generated and processed to improve stitching accuracy in feature-poor areas, then stitching quality improves, but processing time and computational load increase
Solution Approach 1:
The patent performs preliminary actions by generating multiple types of reference images (edge, gradient, Laplacian) before the actual stitching process. These pre-processed reference images contain enhanced feature information that facilitates more accurate and faster feature matching during stitching, reducing the overall processing time despite the additional initial computation.
Solution Approach 2:
The patent applies partial processing by selectively using different types of reference images based on the specific characteristics of the aerial images being processed. Not all reference image types are always generated or used; the system adapts the processing level to the actual needs, balancing accuracy requirements with computational efficiency.
3Adaptability or versatility
If standard single-spectral imaging is used, then the imaging system is simple and fast, but the ability to assess plant growth conditions is insufficient
Solution Approach 1:
The patent implements multi-functionality by using multispectral sensors that can capture images at multiple wavelengths simultaneously. The same imaging system serves both for general aerial photography and for specific plant growth condition assessment by analyzing spectral characteristics at different wavelengths, eliminating the need for separate specialized imaging systems.
Solution Approach 2:
The patent changes the spectral parameter of the imaging system by incorporating multispectral sensors that capture images at multiple wavelengths (e.g., visible light, near-infrared). This parameter change enables the detection of plant physiological states through spectral analysis, providing growth condition assessment capabilities while maintaining a relatively integrated imaging system design.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present technique relates to an image processing apparatus, an image processing method, and a program that can easily execute a stitching process. Provided are: an image generation unit that generates a first reference image regarding a first imaging region on the basis of a plurality of first images regarding the first imaging region and that generates a second reference image regarding a second imaging region at least partially overlapping with the first imaging region on the basis of a plurality of second images regarding the second imaging region; and a processing unit that generates positioning information indicating a correspondence between the first imaging region and the second imaging region on the basis of the first reference image and the second reference image. The present technique can be applied to, for example, an image processing apparatus that executes a stitching process of a plurality of images.