Abstract
Road extraction in rural areas is one of the most fundamental tasks in the practical application of remote sensing. In recent years, sample-driven methods have achieved state-of-the-art performance in road extraction tasks. However, sample-driven methods are prohibitively expensive and laborious, especially when dealing with rural roads with irregular curvature changes, narrow widths, and diverse materials. The template matching method can overcome these difficulties to some extent and achieve impressive road extraction results. This method also has the advantage of the vectorization of road extraction results, but the automation is limited. Straight line sequences can be substituted for curves, and the use of the color space can increase the recognition of roads and nonroads. A model-driven-to-sample-driven road extraction method for rural areas with a much higher degree of automation than existing template matching methods is proposed in this study. Without prior samples, on the basis of the geometric characteristics of narrow and long roads and using the advantages of straight lines instead of curved lines, the road center point extraction model is established through length constraints and gray mean contrast constraints of line sequences, and the extraction of some rural roads is completed through topological connection analysis. In addition, we take the extracted road center point and manual input data as local samples, use the improved line segment histogram to determine the local road direction, and use the panchromatic and hue, saturation, value (HSV) space interactive matching model as the matching measure to complete the road tracking extraction. Experimental results show that, for different types of data and scenarios on the premise, the accuracy and recall rate of the evaluation indicators reach more than 98%, and, compared with other methods, the automation of this algorithm has increased by more than 40%.
Highlights
Road extraction from remote sensing images plays an incredibly important role in map updating, emergency responses, smart cities, sustainable urban expansion, vehicle management, urban planning, traffic navigation, public health, drone navigation, disaster management, agricultural development, driverless vehicle routing, and traffic management [1,2,3,4,5,6]
We combined a model-driven method with a sample-driven method to maximize the degree of automation in road extraction in rural areas on the basis of ensuring the accuracy
To address the low automation degree in rural road extraction caused by prominent curvature differences, complex road paving materials, and narrow road width, we proposed a model-driven-to-sample-driven for rural road extraction method that applies line sequences to curves and connects panchromatic images to HSV space to improve the contrast between road and nonroad areas
Summary
Road extraction from remote sensing images plays an incredibly important role in map updating, emergency responses, smart cities, sustainable urban expansion, vehicle management, urban planning, traffic navigation, public health, drone navigation, disaster management, agricultural development, driverless vehicle routing, and traffic management [1,2,3,4,5,6]. Using remote sensing images to complete road extraction has become popular. For the practical road extraction algorithm using remote sensing image to complete road engineering, the results are as follows: (1) high precision and recall, (2) high degree of automation, and (3) the results can be vectorized. In China, the total mileage of rural roads has reached 4.2 million kilometers, accounting for 83.8%
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
