Title | Method For Estimating Depth Value Of A Two Dimensional Image |
---|---|
Project Leader | |
Name | Rahmita Wirza O. K. Rahmat |
Telephone No. | 038989466595 |
rahmita@fsktm.upm.edu.my | |
Faculty | Computer Science and Information Technology |
IP Information | |
IP Type | Patent |
Filed Date | 07-05-2011 |
Application No. | PI2011003152 |
Country Filing | Malaysia |
Applicant | Universiti Putra Malaysia |
Granted Date | |
IP Number | MY149744A |
Abstract | |
The present invention relates to a method for estimating depth value of a two dimensional image characterized by: first, placing an object on a rotatable plate at a predetermined distance from an image capture; then, acquiring a first image of the object and calculating Scale Factor; then, rotating the rotatable plate to rotate the object to a predetermined angle; then, acquiring a second image of the object; then, obtaining feature point coordinates by applying Good Features to Track technique; then, collecting the coordinates value in the form of (xv1,i, yv1,i) and (xv2,i, yv2,i); then, filtering feature point coordinates which are outside the controlled environment; then, applying Lucas-Kanade Optical Flow technique to the coordinates value obtained from the Good Feature to Track technique to find matching feature point coordinates from the first image and second image; then, calculating Magnitude Optical Vector for each of the feature point coordinates; then, calculating World Coordinate for each of the feature point coordinates; then, calculating World Magnitude Optical Vector for each of the feature point coordinates; then, estimating depth value of the acquired image by calculating each feature point coordinates; then, filtering feature point coordinates using Gaussian distribution method to eliminate noise. | |
Technology | - |
Inventor | Rahmita Wirza, Suhail Azmi, Ng Seng Beng |
Statement of Novelty | |
Work Start Date | 01-01-0001 |
Work End Date | 01-01-0001 |
Classification | |
Commercialised Status | |
Faculty | Computer Science and Information Technology |