Real-Time Linear Feature Detection in Aerial Imagery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional visual inspection systems for linear infrastructural objects, such as power lines, face challenges in real-time detection due to high computational complexity and limitations in handling varying backgrounds, especially in outdoor scenes, where existing techniques like Hough transform and canny edge detection are not robust enough for near-real-time processing and often result in false positives.
Innovation Solution
A computer-implemented method and system that applies background suppression techniques, converts HSV images to grayscale and binary, uses mean shift filtering, generates gradient images with the Sobel function, extracts seed point pairs along linear features, and employs a contour growing approach to detect boundaries, while removing false positives based on rigidity features and total gradient orientations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Hough transform is used for linear feature detection, then detection capability is improved, but computational complexity increases to O(n³) making it unsuitable for real-time processing
Solution Approach 1:
The image processing is segmented into multiple stages: background suppression using HSV color space, edge detection using Sobel operator, line segment extraction, and Hough transform application only on extracted segments. This segmentation reduces the input size for the computationally intensive Hough transform from the entire image to only relevant line segments, achieving real-time processing while maintaining detection accuracy.
2Measurement precision
If canny edge detection is used for linear feature detection, then detection capability is improved, but computational complexity increases making it unsuitable for near-real-time processing
Solution Approach 1:
Instead of applying full Canny edge detection to the entire image, the patent applies Sobel operator for gradient calculation only after background suppression, and only processes regions where edges are likely to exist. This partial application of edge detection maintains sufficient accuracy for power line detection while dramatically reducing computational load to achieve near-real-time processing.
3Device complexity
If background subtraction is used for outdoor scenes, then processing is simplified, but detection accuracy decreases due to non-stochastic nature of outdoor backgrounds
Solution Approach 1:
The patent transforms the image from RGB color space to HSV (Hue, Saturation, Value) color space, which changes the parameter representation of colors. This transformation makes the background suppression more effective because outdoor backgrounds (sky, greenery) have distinct HSV characteristics that can be suppressed while preserving power line features, achieving both simplified processing and improved detection accuracy.
4Measurement precision
If piloted aircraft are used for visual inspection, then inspection quality is improved, but cost and risk increase significantly
Solution Approach 1:
The system uses automated image processing algorithms (background suppression, edge detection, Hough transform) that process images autonomously without requiring human operators to manually inspect each image. The computer vision system performs the inspection task itself, eliminating the need for expensive piloted aircraft operations while maintaining high detection quality through automated analysis.
Data Source
AI summary
This disclosure relates generally to visual inspection systems, and more particularly to a method and system for facilitating real time detection of linear infrastructural objects in aerial imagery. In one embodiment, a background suppression technique is applied to one or more hardware processors to a HSV image. Further, a mean shift filtering technique is applied to the hardware processors to find a peak of a confidence map and then a gradient image generation is performed for a plurality of edges of the image. A seed point pair along a middle cut portion of a linear feature of the HSV image to identify one or more boundaries of the seed point pair is extracted and then a contour growing approach to detect the boundaries of the linear feature is initiated. Lastly, one or more false positives are removed by using a rigidity feature, the rigidity feature being equivalent to the total sum of gradient orientations.