Adaptive Segmentation Evaluation.

Abstract

Phase 2 of this contract addressed the issue of improving the segmenter performance on military vehicles in infrared imagery through the use of temporal processing techniques. Specific objectives were: 1) Develop temporal-based techniques to augment current segmentation algorithms; 2) Develop a set of metrics to quantitatively represent segmenter performance in terms of quality and consistency of segmentation; and 3) Perform a comparative study of performance results between the modified segmentation approach and the unaided approach. The approach concentrated on developing techniques for image data stabilization through the use of multiframe data integration. The most important aspect of multiframe data integration is its ability to increase frame to frame data stability and reduce noise. These properties have two important consequences: 1) the improved signal quality greatly reduces the need for special purpose processing by each ATR component to overcome the image ambiguities found in the raw data; 2) features that represent higher levels of structural detail usually masked by noise can be computed for improved object discrimination and classification performance.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 24, 1987
Accession Number
ADA190965

Entities

People

  • Ron Patton

Organizations

  • Martin Marietta

Tags

Communities of Interest

  • Materials and Manufacturing Processes
  • Sensors

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Classification
  • Computer Vision
  • Contracts
  • Data Integration
  • Detection
  • Detectors
  • Filters
  • Image Processing
  • Military Vehicles
  • Plastic Explosives
  • Sequences
  • Target Recognition
  • Test Sets
  • Three Dimensional
  • Vehicles

Readers

  • Computer Vision.
  • Systems Analysis and Design