Hand Gesture Optimization Using Structure From Motion

Muhammad Sonnie Bestian, Risanuri Hidayat, I Wayan Mustika

Abstract


Utilization human body with unique way to distinguish between one person and another already widely used in research. Biometric is a method for recognizing humans based on one or more physical characteristics or unique behaviors. Many previous researches of hand gestures used to control devices and identification. In This paper, we discusses the use of structure from motion method for optimization hand gesture. Structure from Motion ( SFM ) is the process to estimating the structure of 3-D shape from the series of 2-D images. The goal is to estimate the pose of the camera calibration results bits of the image to be reconstructed into a 3-D structure, from the scene and the scale factor is not known, then recover the actual scale factor by displaying an object of known size. The results from the trial by reconstructing four (4) drawing hand pieces and matched by detecting a hand gesture (3-D), from the stage of the process generated frame 2 + 3: match rate slightly (0) => relax SIFT threshold to 0.7 to 0 matches => SIFT relax the threshold to 0.8 with 2 matches => SIFT relax the threshold to 0.9 with 18 matches => SIFT relax threshold to 0.95 with 42 matches. From the test results obtained by the degree of fit pieces of the picture (2-D) to the whole image (3-D) resulted in 53,85% accuracy value.




DOI: 10.25037/pancaran.v7i2.172

Refbacks

  • There are currently no refbacks.


Copyright (c) 2018 Pancaran Pendidikan