Digital image processing helps replace several mundane activities. My project in final year was ‘Application Control using Hand Gesture Recognition From 3-Dimensional Images’. We worked extensively on processing a 3-D image to recognize the encrypted gesture, with the added 3rd dimension for more number of gestures. We started with extensive image segmentation to suppress background, handle dynamic lighting conditions and extract only portion of hand from the 3-D image. A tedious task, we researched many papers on IEEE and books on Image processing, to come-up with a complex code logic involving a combination of three different segmentation techniques involving RGB, YCbCr and HSV models. Further, with mathematical calculations involving use of Eigen values and Eigen vectors were derived based on the co-variance matrix, generated as per captured image, we managed to extract maximum information. A Euclidean distance was calculated to denote the deviation of captured image from each set of pre-captured images for all defined gestures. Hectic code optimization helped save precious execution time. The gesture corresponding to minimal Euclidean distance was the identified gesture. This project acquainted me with MATLAB and I learnt about various image enhancement and image segmentation techniques. I presented a paper at IEEE’s International Conference on Convergence of Technology (I2CT), Pune, published in the journal with ISBN “978-1-4799-3759-2”.
Digital image processing helps replace several mundane activities. My project in final year was ‘Application Control using Hand Gesture Recognition From 3-Dimensional Images’. We worked extensively on processing a 3-D image to recognize the encrypted gesture, with the added 3rd dimension for more number of gestures. We started with extensive image segmentation to suppress background, handle dynamic lighting conditions and extract only portion of hand from the 3-D image. A tedious task, we researched many papers on IEEE and books on Image processing, to come-up with a complex code logic involving a combination of three different segmentation techniques involving RGB, YCbCr and HSV models. Further, with mathematical calculations involving use of Eigen values and Eigen vectors were derived based on the co-variance matrix, generated as per captured image, we managed to extract maximum information. A Euclidean distance was calculated to denote the deviation of captured image from each set of pre-captured images for all defined gestures. Hectic code optimization helped save precious execution time. The gesture corresponding to minimal Euclidean distance was the identified gesture. This project acquainted me with MATLAB and I learnt about various image enhancement and image segmentation techniques. I presented a paper at IEEE’s International Conference on Convergence of Technology (I2CT), Pune, published in the journal with ISBN “978-1-4799-3759-2”.