NEURAL FUZZY LOGIC FOR AUTOMATIC OBJECT RECOGNITION

1
NEURAL - FUZZY LOGIC FOR AUTOMATIC OBJECT
RECOGNITION

• .

2
INTRODUCTION

• Artificial Neural Networks is a system modeled on
  the human brain. It is an attempt to simulate
  with specialized hardware or sophisticated
  software the multiple layers of simple processing
  elements called neurons.
• Fuzzy set theory resembles human reasoning in its
  use of approximate or corrupted data to generate
  decision.
• Integration of ANN and Fuzzy logic can provide a
  very efficient solution for Target recognition.

3
MODULES OF THE ATR SYSTEM

• Acquisition Capturing data with a sensor.
• Transformation Preprocessing of the image.
• Segmentation Identifying regions of interest,
  using Freeman code for border detection.
• Geometric features Usage of Hough Transform for
  identifying hidden lines also.
• Target Database Tabulated values for each model.
• Model Matching Matching measured values with
  tabulated values.

4
IMPLEMENTATION OF NEURAL - FUZZY LOGIC

• With the inclusion of fuzzy logic in the ATR
  system, even when one dimension is obscured a
  match can be made with the remaining two
  dimensions.
• Network is trained using a Back Propagation
  algorithm

5
BACK PROPAGATION ALGORITHM
6
OTHER APPLICATIONS

• Character Recognition
• Automatic Phonetic Recognition
• Facial Recognition
• Signature Recognition
• Fingerprint Recognition

7
CONCLUSION

• The computing world has a lot to gain from Neural
  Networks.
• Their ability to learn makes them flexible and
  very powerful.
• The most exciting aspect of Neural Networks is
  the possibility that some day conscious
  networks may be produced.
• Neural Networks have a huge potential and we will
  get the best of them when integrated with
  computing, AI, Fuzzy logic and related subjects.

8
BIBLIOGRAPHY
9
PRESENTED BY

• NAGINI INDUGULA (4/4 CSE) 98311A0515

Sree Nidhi Institute of Science and Technology

Hyderabad
10
?? QUESTIONS ??