Jasmine Bhanushali

Email: jbhanush-at-uci.edu


My work is in the area of Image Processing,Computer Vision and Machine Learning.


CT-X-Ray Registration
In this project, I had a dataset of images of CT of the lumbar spine and X-Rays of the spine taken at different angles. I generated the digitally reconstructed radiograph from the CT data and then measured the similarity with the X-Ray data to determine the angle. This is useful as CT scan cannot be performed during an operation or during radiation therapy. And hence intraoperative X-Ray is compared with a Digitally Reconstructed Radiograph from pre-operative CT to correctly determine the position.
Removing Eye-glasses from Photos
We identified the region occluded by glasses in an image and suitably replaced it to generate a natural looking glassless image of the face. We first generated the face using Eigen Face reconstruction and then found the approximate occlusion region. We then filled the occluded region in using the method of recursive reconstruction


  • Clear Guide Medical (2016) : I was an intern for the summer at Clear Guide Medical. I worked on different methods of segmentation of markers on CT, CBCT and MRI Volumes. I then worked on segmentation of translucent markers from the video stream and then visualized them in 3-D with respect to the camera position. This made a significant contribution towards adding the MRI modality for the fusion system.
  • Smarttrak Solar Systems (2015-16) : I worked as an embedded design enginer at Smarttrak. I worked on the wireless communication system and also on the FPGA-based design of the Solar tracker.
  • Head Teaching Assistant (2015) : I was the Head Teaching Assistant in Spring 2015 and was responsible to ensure that the tutorials and labs were carried out properly. I was involved in managing 11 other TAs and was also responsible for taking tutorials and correcting assignments.


A Dome-Shaped Interface Embedded with Low-Cost Infrared Sensors for Car-Game Control by Gesture Recognition
Human Computer Interaction Conference 2015, Los Angeles
Jasmine Bhanushali, Sai Parthasarathy Miduthuri, Kavita Vemuri
We made a gesture controller which could identify complicated gestures like turning and twisting gestures using only IR sensors. 9 IR sensors were placed at strategic locations on a dome and the data received was analyzed. Hidden Markov Model was then used to accurately predict the gesture performed.


A copy (updated on Dec 2016) is available here.


  • Email: