Hand gesture controlled

Virtual Earth




Objective


The objective of this project was to develop a system that is capable to interact with the system and is specifically based for those devices that are not touch-based.

A device equipped with webcam or external camera can be used to receive images and process them to determine user input via sequence of images. This project is an image processing and Mono-Vision based project, a system capable od detecting objects in the 3D real world and extract information based on their color. The system tries to detect hand based on color segmentation in HSV format. The detected object is traced and tracked to control a 3D motion of object in virtual environment.




System side concept flow


  • Webcam Input Image
  • Image segmentation
  • Determine the countor
  • Determine the hand center
  • Track the hand



Design and Development hurdles




image

The background coincides with skin color

image

Poor source of illumination

image

System dependence can lead to slow processing

image

Noise in the image frames

image

Variation in skin color

image

Presence of directional light






Information flow


  • CaptureFrame one frame at a time
  • Convert the image from RGB to HSV
  • Apply HSV based segmentation
  • Apply Edge Detection algorithm
  • Find Edge points for each of the encounters
  • Fill the space of each contour
  • Find space of each contour
  • Find the contour with maximum area
  • Find the centroid of largest area



Final working prototype




image image





© 2020 Noopur Kumari Kashyap