Optimizing storage for real-time road monitoring with dual-camera systems

Abstract

Efficient and accurate road surface distress detection is essential for enhancing road safety and optimizing maintenance strategies. A significant challenge in pavement distress detection lies in developing an optimized data collection approach. This paper introduces a real-time road surface distress detection system utilizing a dual-camera setup mounted on a vehicle. The system incorporates an RGB camera at the front and an RGB-D (stereo vision) camera at the rear, both capturing videos stored in temporary buffers. When the front camera detects anomalies, e.g., cracks or potholes, the corresponding frames from both cameras are saved. Depth-enabled frames from the rear camera facilitate precise assessment of distress dimensions. To optimize storage, only event-related frames are retained along with pre- and post-event context. This selective strategy significantly reduces storage requirements while preserving critical data for analysis. Simulations validate the system’s efficiency, demonstrating substantial storage savings compared to continuous recording methods while maintaining effective road condition monitoring.