Skip to main content

Car Detection by Fusion of HOG and Causal MRF

You are here

Detection of cars has a high variety of civil and military applications, e.g., transportation control, traffic monitoring or surveillance. It forms an important aspect in the deployment of autonomous Unmanned Aerial System (UAS) in rescue or surveillance missions. In this paper, we present a two-stage algorithm for detecting automobiles in aerial digital images. In the first stage, a feature based detection, based on local histogram of oriented gradients (HOG) and support vector machine (SVM) classification, is performed. Next, a generative statistical model is used to generate a ranking for each patch. The ranking can be used as a measure of confidence or threshold to eliminate those patches that are least likely to be an automobile. We analyze the results obtained from three different types of data sets. In various experiments we present the performance improvement of this approach compared to a discriminative- only approach, e.g., the false alarm rate is reduced by a factor of seven with only ten percent drop in the recall rate.

Contact: 
Satish Madhogaria

Field of Interest

“The field of interest shall be the organization, systems engineering, design, development, integration, and operation of complex systems for space, air, ocean, or ground environments. These systems include but are not limited to navigation, avionics, mobile electric power and electronics, radar, sonar, telemetry, military, law-enforcement, automatic test, simulators, and command and control."

Resources

Technology Navigator
Spectrum

Send Us A Message