Contact us for more information about our business solutions and capabilities
 

spacer
About CACI CACI Careers CACI Investor Info Contact CACI spacer
Search
spacer
blackline
TWIST Image Registration and Change Detection Tool
Proven Industry Performance.

Find out more about how CACI MTL can help you!

What TWIST Does
  • Automatically registers (spatially aligns) two images in translation, rotation and size with sub-pixel accuracy
  • Automatically detects differences in two images due to scene or object changes
  • Can be tuned to detect selected types of differences (vehicles, structures, etc.)
TWIST Applications
  • Detecting object changes or relocations, documentation changes, manufacturing defects, metal fatigue, water quality changes, tumor growth, population growth, etc.
  • Analyzing large numbers of images rapidly and accurately
  • Detecting image differences in real time for time critical targets and/or rapidly changing conditions

 

CACI MTL's TWIST is Unique
  • Advances the state-of-the-art in processing speed, accuracy, versatility and storage efficiency
  • Can register images from different types of sensors (visual, infrared, radar, etc.) for data fusion
  • Two-axis image sorting technique uses CACI MTL's proven statistical signal processing (SSP) technology to holistically capture all relevant image features with sub-pixel accuracy and orders of magnitude less processing time and storage than traditional techniques
How TWIST Works
  • Treats each row and column of pixels in an image as a spatially varying signal (pixel intensity vs. location)
  • Registers images and detects image changes based on spatial variations in image complexity (pixel intensity and frequency)
  • Computes row and column axis functions (AFs) as complexity metrics and compares them for each image to perform registration (global min differences in AFs) and change detection (local max difference in AFs)

Peak Values of Image Difference
Metric Indicate Cell Locations of
Image Differences

TWIST Features TWIST Benefits
Speed - orders of magnitude faster than traditional techniques, which extract many more features (polygons, lines, points, etc.) and perform more complex processing (splines, genetic algorithms, etc.) High Speed - makes on-the-fly, image registration and difference detection practical
Accuracy - capable of sub-pixel accuracy in automatically identifying image differences Enhances Effectiveness - of operator/analyst by reducing errors caused by fatigue and boredom
Storage - requires orders of magnitude less storage than traditional techniques Simple, Efficient Storage - is ideal for low cost, remote sensing and processing
Input Data - images from multiple types of sensors, including photographic, video, laser, radar, infrared, multispectral, hyperspectral, polarimetric, acoustic, biometric, X-ray, MRI, etc. Sensors can be at different viewing angles or distances from object. Versatility - allows registration and difference detection of object images sensed by different sensor types at different angles and distances

TWIST H/W-S/W Configuration - PC-based under any Windows O/S, written in C++ and Visual Basic

Original Sponsor - CACI MTL developed the TWIST technology under IR&D and applied it for a classified U.S. government customer

Download brochure PowerPoint logo PowerPoint - 422 KB

red arrow See the demonstration video