Chest boundary shape estimation using a RGBD camera and pattern markers for robust lung electrical impedance tomography

We propose a novel chest boundary estimation method using a RGBD camera for robust lung electric impedance tomography (EIT). For EIT imaging, sixteen electrodes are generally used, so we use the twelve electrodes as color markers and put pattern markers on the rest four electrodes. The color and pattern markers are detect to extract surface patches. The detected pattern markers are refined by affine parameters to correctly recognize the patterns. The boundary shapes are reconstructed by the registration using translation and rotation of consecutive surface patches. In the experiment results, the relative error of a round table is 0.0169, the processing fame rate is more than 12 frames/sec without GPU (graphics processing unit), and the more robust EIT reconstruction results are shown.