39 Electrocardiogram (ECG) Image processing and Extraction of Numerical Information International Journal of Engineering Technology Science and Research IJETSR www.ijetsr.com ISSN 2394 – 3386 Volume 3, Issue 7 July 2016 Electrocardiogram (ECG) Image processing and Extraction of Numerical Information Dharmendra Gurve 1 , Alok Kumar Srivastava 2 , Kingsuk Mukhopadhyay 2 , N Eswara Prasad 2 , Sachin Shukla 1 , H. Muthurajan 1 1 National Center for Nano Science and Nano Technology, University of Mumbai, Vidyanagri, Kalina, Santacruz(E), Mumbai 2 Defence Materials Research & Development Establishment (DMSRDE), DRDO GT Road Kanpur, India Abstract: An Electrocardiogram (ECG) is a graphic tracing of electrical patterns produced by the heart. This test is frequently used for patients who have heart problems and is an important diagnostic procedure. Standard ECG are recorded on paper has a grid on it. The grid work on ECG records is made up of many tiny blocks that are 1 mm square. Each of these tiny boxes is generally 40 ms duration. Heavier lines are used to make larger squares of five boxes tall and five boxes wide. Each larger square has a duration of 200 ms. Five of these larger squares (200 × 5) equals 1000 ms or 1 sec. By counting out the grids, we can get a fast approximation of the duration of a particular cardiac cycle or timing cycle. However this becomes increasingly complicated when we would like to analysis the ECG data for higher precision. The normal ECG record comprises a P wave, QRS complex, ST segment, and T wave. Analysis of each segment duration is very important for critical decisions by clinician. In view of this we have developed MATLAB based algorithm to read the ECG images, which has options to calibration in both X and Y axis, followed by algorithm to remove the background grids and extract the numerical co-ordinates of individual points of ECG curve. The numerical data extracted from ECG are highly useful for high precision diagnosis and the details are discussed in the manuscript. Keywords: Image Processing, Electrocardiogram (ECG), Background elimination, Image. 1. Introduction Analysis of electrocardiogram (ECG) as a tool for clinical diagnosis has been an active research area in the past decades. The validity of using ECG tracing for heart disease detection supported by the fact that the physiological differences of the heart in different individuals display certain uniqueness in their ECG signals [1]. Human individuals present different patterns in their ECG regarding wave shape, amplitude, PT interval, due to the difference in the physical conditions of the heart [2]. Further, ECG signal is a life indicator, and can be used as a tool for liveness detection. An ECG graph paper can find wide applications in physical access control, medical records management and forensic applications. ECG tracing is printed on graph papers for accurate interpretation. On ECG graph the Y-axis represents voltage in mV. The bar indicated below represents 1 millivolt (mV). This denotes the electrical strength of the signal. Therefore, each big square represent 0.5 mV and each small square represent 0.1mV. The X-axis represents time in seconds. Each big square represent 0.2 seconds. Therefore, each small square represents 0.04 seconds. There any application of ECG tracing on graph paper but in many application ECG tracing without grids are needed, so in such cases extraction of plane ECG signal without grids from the scanned graph paper is important. This process is known as feature extraction or image background removal. Background subtraction is a process of extracting foreground objects in a particular image. The foreground object boundaries extraction reduces the amount of data to be processed and also provide important information about the object. The various methods are presented to remove background, among them (SHARP) is the widely used background field removal technique [3]. SHARP uses the spherical mean value (SMV) property to remove the harmonic component, i.e., background field, from the measured total field. Pixel-based techniques assume that the time series of observations is independent at each pixel. In contrast, some researchers [4-6] employ a region- or frame based approach by segmenting an image into regions or by refining low-level classification obtained at the pixel level. Markov random field