Calibration Of EEG Systems By Computer Vision

EEG (electroencephalogram) provides a great help for clinical doctors in the diagnosis of brain diseases. In particular, it has irreplaceable value in the diagnosis of epilepsy. Determination of the position of brain source is important, and it is more important to estimate the positions of EEG electrodes.In recent years, researchers in various fields have put forward five kinds of method to calculate the positions of EEG measuring electrodes. Those five methods have achieved good results. They are:(i) Manual methods,(ii) Electromagnetic radiofrequency digitizer usage,(iii) Magnetic Resonance Imaging (MRI)-assisted approaches,(iv) Ultrasonic wave transmission and reflection methods,(v) Photogrammetric measurement principles. With the development of computer vision, development of photogrammetry method has made a great progress. Compare to the other measurement ways, photogrammetry is more fast, convenient and low cost. Most important is that the brain source is electronic signal. In the environment of other methods, brain source can be affected by electromagnetic interference. This problem can be avoid by photogrammetric measurement. In the several different methods of Photogrammetry, each have advantages and disadvantages. Aiming at the application background, thorough research has been made in photogrammetry measurement method for determining the positions of EEG electrodes in brain signal acquisition system. Existing methods collect image data mainly through monocular vision principle, principle of binocular vision and visual principle to and to localize and identify the EEG electrodes. In practical, speed, cost, accuracy, convenient always cannot be satisfied simultaneously. Either the cost is relatively high, or need a lot of auxiliary facilities to ensure the speed and accuracy at the same time. This paper proposes a new photogrammetry framework for determining the positions of EEG electrodes. The main work and results are as follows:1. This paper proposes a novel photogrammetry framework for fast localizing the EEG Sensors based on Microsoft Kinect equipment. Because of the depth sensor can obtain the3D information directly. So we can calculate the3D coordinates of the EEG electrodes in each single view in this photogrammetric system with the depth sensor equipment. In order to obtain all of EEG electrodes, this paper capture data in four different views.2. This paper proposes a local repair algorithm for Kinect depth map. This paper uses Microsoft Kinect to collect data. We could know that the error of the depth map is about3mm when the shooting distance is0.8m-1.2m after analyzing the noise of Kinect depth map and the character of Kinect. Thus, we improved bilateral filtering function. And it greatly improves the filtering speed. Deletion of Kinect depth map is more serious. It cannot achieve good repair effect only by filtering. Considering the integrity of RGB images, we fill the depth missing pixels by combining the texture. That implements through the SAD matching principle of stereo vision. Finally, we achieve good repair results.3. This paper designed a calibration process for proposed photogrammetry framework for fast localizing the EEG electrodes. In the previous photogrammetry methods on localization of EEG electrodes, principle of stereo vision has been used to calculate the three-dimensional coordinates of electrodes directly. The Kinect can obtain the3D information of scene. So the3D location of the electrodes can be achieved from a single view. But these EEG Sensors are distributed among four different coordinates. Therefore, a calibration method is proposed which can obtain the conversion relationship among those four perspectives to enhance the practicability of this system.Finally, this dissertation verifies the proposed method and implements the detection and localization of EEG electrodes. The simulation experiments were done through a plastic head with black marks which represents the EEG electrodes. We implement the identification and localization of EEG electrodes by using our proposed photogrammetric system and the experimental results demonstrate that our proposed photogrammetry framework for fast localizing EEG electrodes is practical.