aligning a raw image to a real time coordinate system on the web
Download
Skip this Video
Download Presentation
ALIGNING A RAW IMAGE TO A REAL TIME COORDINATE SYSTEM On THE WEB

Loading in 2 Seconds...

play fullscreen
1 / 23

ALIGNING A RAW IMAGE TO A REAL TIME COORDINATE SYSTEM On THE WEB - PowerPoint PPT Presentation


  • 112 Views
  • Uploaded on

Aniket Phatak UNI: avp2110. ALIGNING A RAW IMAGE TO A REAL TIME COORDINATE SYSTEM On THE WEB. Image Search Engine Results now Focus on GIS image registration The Technique and its advantages Internal working Sample Results Applicable to other areas like face recognition etc. Future scope.

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about ' ALIGNING A RAW IMAGE TO A REAL TIME COORDINATE SYSTEM On THE WEB' - ova


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
topics covered

Image Search Engine Results now

Focus on GIS image registration

The Technique and its advantages

Internal working

Sample Results

Applicable to other areas like face recognition etc.

Future scope

TOPICS COVERED
current image search technique

Image search is a complex and costly task

Hence, present web search engines query the title or the metadata of the image to get results faster.

Adversarial attack is the huge problem associated with above technique

Hence, we need to devise some algorithm that can some significant pixels in image for image comparison

Focus now is on Web-based georeferencing.

Current image search technique
geographic images on web

Google Earth and corresponding maps.google.com has set high standards for all web applications and websites dealing with high resolution/high accuracy geographical feature content. It can be used using APIs.

  • The programming environment of Flex SDK and corresponding scripting language Actionscript v3.0 embedded in Adobe Flash CS3 has enabled the use of Google Maps library in Flash Applications.
  • Required for this:-
    • High Internet speeds
    • Geo-referencing
Geographic Images on web
holistic view of process

Using Principal Component Analysis(PCA) technique, the most similar image from the database is selected.

Now some specific significant pixels named Control Point Pairs(CPPs) are selected for image registration automatically.

Next time, for image registration and georeferencing on any other server, we just need to pass these CPPs instead of whole image.

Holistic view of process
introduction
INTRODUCTION

Raw Image

Real World Map

Aligning a raw image with a real world map coordinate system.

geo referencing process
GEO-REFERENCING Process

Spatial datasets from different sources need to be accurately aligned geographically in order to be viewed or analyzed together

georeferencing process

Georeferencing is one of the vital research areas of GIS data integration literature. Geospatial information needs to be extracted from multiple sources in a very consistent and precise way. The typical Georeferencing process includes:

  • Identifying a set of control point pairs that link locations on a raster image with corresponding locations on a correctly positioned vector dataset.
  • Calculating a transformation function from a raster image to the vector map based on the Control Point Pairs (CPPs).
  • Transforming and re-sampling the image.
Georeferencing process
issues

Manually Finding CPPs is

  • Time consuming
  • Tedious
  • Sometimes impossible
  • Must know a priori approximate location
  • Distorted and Transformed images makes it even harder to identify the location.
ISSUES
solution

AUTOMATED GEO-REFERENCING

  • Requires no pre knowledge of the image’s placement in the road network.
  • Necessitates only a few points from the image.
  • Tolerates point location distortion , missing points and spurious points
  • Provides high performance and scalability
SOLUTION
image enhancement

Process by which an image is manipulated to increase the amount of information perceivable by the human eye.

  • Inputs: neighborhood pixels, intensity, gray level values .
  • Outputs: enhanced (smoothened) image .
  • Algorithms : delta-connected components, symmetric neighborhood filters .
IMAGE ENHANCEMENT
image segmentation

Process of partitioning the image into non overlapping regions according to gray level, texture etc

  • Single priority queue
IMAGE SEGMENTATION
image registration

Process of overlaying two or more images of the same scene taken at different times, from different from different view points.

  • Geometric alignment of images.
  • Correlation function used for feature matching.
  • Comprises of:
  • Feature detection.
  • Feature matching.
  • Transformations.
IMAGE REGISTRATION
image registration1
IMAGE REGISTRATION

Image Registration Algo

DATABASE REFRENCE IMAGE

INPUT IMAGE

image registration and transformations
IMAGE REGISTRATION AND TRANSFORMATIONS

PIECEWISE LINEAR

AFFINE TRANSFORM

INPUT IMAGE

PROJECTIVETRANSFORM

LWM TRANSFORM

limitations

Only spatial datasets from different sources are considered.

  • A minimum of 4 control points in image is required for matching.
  • Pattern matching is currently only being done on point data.
LIMITATIONS
future scope

Easily extended to other image matching applications like face recognition etc.

  • Natural Disaster management.
  • Implementing GIS Applications and Pattern Matching for paleontological classification of ammonitic suture.
  • Housing Stock surveys.
FUTURE SCOPE
conclusion

An image search engine can use this algorithm to avoid storing various copies of same image location.

  • It can register images from different sources and align them without actually comparing them pixel by pixel each time which is time consuming and costly process.
  • Easily scalable architecture and more suitable for distributed environment where network bandwidth is precious.
  • Removes manual human intervention and thereby any possibility of human error in image matching.
CONCLUSION
ad