Segmented Image Classification for Remote Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated methods for assessing object characteristics from digital images in remote sensing are computationally expensive and provide limited information, falling short of the accuracy and detail offered by pixel-wise classification while being less resource-intensive than fully convolutional neural networks.
Innovation Solution
The method involves subdividing digital images into segments, using General Image Classification to assess each segment independently, and determining the confidence level of predetermined characteristics, producing a generalized label with spatial rejoining for enhanced mapping and reduced computational overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pixel-wise classification (fully convolutional neural networks) is used to assess each pixel individually, then measurement precision and detail information are improved, but computational resources and processing time increase significantly
Solution Approach 1:
The patent divides the digital image into multiple segments or regions, and applies classification algorithms to each segment rather than processing every individual pixel. This segmentation approach maintains detailed information assessment while significantly reducing computational resources by grouping pixels into manageable segments that can be processed independently and efficiently.
2Device complexity
If General Image Classification is used to assess image segments, then computational resources are reduced, but measurement precision and detail information decrease
Solution Approach 1:
The patent applies different processing approaches to different segments of the image based on their specific characteristics and importance. Critical regions receive more detailed analysis while less critical areas use simplified classification, optimizing the balance between computational efficiency and measurement precision across the entire image.
Solution Approach 2:
The patent transitions from pixel-level (2D) classification to segment-level (coarser 2D or hierarchical) classification, adding a spatial dimension to the analysis. By rejoining assessed segments to form generalized labels, the system recovers detailed spatial information while maintaining the computational efficiency of segment-based processing.
3Productivity
If automated assessment methods are implemented, then productivity and time efficiency are improved, but measurement precision compared to manual inspection decreases
Solution Approach 1:
The patent replaces manual inspection mechanisms with automated computer-based image classification systems. By using trained classification algorithms that have learned from extensive training data, the system achieves automated assessment with accuracy comparable to or exceeding manual inspection, while dramatically improving productivity and eliminating human factors such as fatigue or subjectivity.
Data Source
AI summary
Methods and systems for automatic estimation of object characteristics from a digital image are disclosed, including a method comprising sub-dividing into two or more segments a digital image comprising pixels and depicting an object of interest, wherein each segment comprises two or more pixels; assessing content depicted in one or more of the segments for a predetermined object characteristic using machine learning techniques comprising General Image Classification of the one or more segments using a convolutional neural network, wherein the General Image Classification comprises analyzing the segment as a whole and outputting a general classification for the segment as a whole as related to the one or more predetermined object characteristic; and determining a level of confidence of one or more of the segments having the one or more predetermined object characteristic based on the General Image Classification assessment.