SIDE WINDOW DETECTION IN NEAR-INFRARED IMAGES UTILIZING MACHINE LEARNING

摘要

Methods, systems and processor-readable media for side window detection in near-infrared (NIR) images utilizing machine learning. An image-capturing unit can capture an image/video in a near-infrared (NIR) band via a side window of an incoming vehicle. A deformable part model can be generated utilizing a side window detection and B-frame detection in order to obtain a set of candidate side-windows. Side window detection can be performed based on a mixture of a tree model and a shared pool and can be globally optimized with dynamic programming and still-capture to detect the backseat side window boundary utilizing a B-pillar. A false alarm with respect to the deformable part model can be removed utilizing a super pixel generation and a longest-line detection unit in order to generate a refined deformable part model.

主权项

1. A side window detection method, comprising:generating a deformable part model with respect to an image of a vehicle captured in a near-infrared band utilizing a side window detection and a B-frame detection module in order to obtain a set of candidate side-windows;
generating a refined deformable part model utilizing a super pixel generation and a longest-line detection in order to remove a false alarm with respect to said deformable part model; and refining said detection performance of said refined deformable part model based on global regional information utilizing a local self-similarity based metric.