Combines image statistics, convolutional features and trained classification models to automatically detect obstructions and vehicle direction, enabling mobile monitoring without requiring a fixed surveillance platform.

‍

Vehicle-mounted camera systems need to reliably identify road or scene obstructions and determine vehicle movement direction despite changing lighting, backgrounds, viewpoints and other visual conditions.

‍

Provides a vehicle-mounted camera monitoring system that extracts image features, including histogram, statistical and convolutional features, and applies trained classification models to detect scene obstructions and determine vehicle movement direction.

‍

Combines statistical histogram features, convolution/grid features and trained classification models to detect obstructions and determine vehicle direction from a mobile, vehicle-mounted camera platform.

‍