DESIGN AND IMPLEMENTATION OF OBJECT RECOGNITION AND TRACKING SYSTEM USING SCALE INVARIANT FEATURE TRANSFORM ̉̉ K NEAREST NEIGHBORS METHOD WITH PID CONTROL FOR HUMANOID ROBOT
Computer vision is a technology intended to replace the visual function in humans with extracting information and features from an image and analyzing the information. In this thesis report presented the process of design and implementation of object tracking system that was built starting with obje...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/26557 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Computer vision is a technology intended to replace the visual function in humans with extracting information and features from an image and analyzing the information. In this thesis report presented the process of design and implementation of object tracking system that was built starting with object recognition in the early stages and then equipped with yaw and pitch system for tracking the position of the object. SIFT algorithm is used as a feature extraction, KNN is used as a classifier and for estimation of homographic changes in objects using RANSAC. In order for objects to be tracked automatically, PID control is used to correct the coordinates obtained when object recognition with center coordinates of the frame. <br />
<br />
The result of this study indicates that the system successfully recognizes objects with keypoints parameter as feature object and different recognition distance on each object. Recognizable objects in this study consisted of drink, towel, medicine and remote with number of keypoints on each object that is 140, 500, 78, 91. The object can be recognize with a distance of 20-80 cm, the furthest distance on a recognizable object is the towel object and the nearest distance is the remote. Object tracking with PID controller successfully implemented on single object displacements and dynamical displacements on moving objects with 11,7% overshoot value of yaw movement and 0% overshoot value of pitch movement and 1,2 seconds risetime. Object position of x and y-axis can be known based on direction move of servo motor yaw and pitch with 150 degree of servo motor reference, object position in right and below of robot if the degree of servo motors less than 150 degrees, in the other hands, object position in left and above of robot if the degree of servo motors more than 150 degrees. <br />
|
|---|