Digital Signal Processing and Processor

Published i n:   Signal Processing Systems (SiPS), 2015 IEEE Workshop Link: https://ieeexplore.ieee.org/document/7344997/ Topic- Denoising and Baseline Correction of ECG Signals using Sparse Representation In the above paper, the authors have documented of de- noising of   thinly dispersed ECG signal and baseline correction approach.Thinly dispersed and superfluous  representation is a signal processing method which can be used effectively and can be implemented to learn the inner structure from given noisy input signals. The characteristics of various waves P- and T- waves and the QRS complexes which are present in the ECG signals and also the smooth varying baseline wandering which are  unknown   information  in the learned dictionaries of the above-mentioned denoising method. this learned dictionary is used to denoise the ECG signal and also and effectively reconstruct using specific atoms in the dictionary and remove the basel...

Pub. no.: US 2015/0018702 A1 The patent contains information about the apparatus and methods to use this apparatus for denoising the electrocardiograms signal (ECGs) where the percentage of amount of filter applied to the signal can be varied manually by the operator or automatically by the program. For the denoising of the signal filter techniques such as filter to correct baseline wander and Savitzky-Golay denoising technique and threshold smoothing is used. The denoising method given in the patent can be described as Firstly providing ECG signal to the computing device and applying the filter to the signal. These filtered signal is displayed on the computing device after which the amount of the filter can be adjusted by the medical professional or automatically by the computing device as per the set mode. Here the amount of filtered is applied in real time as the signal is being displayed. This method provided in patent can be used for processing of ECG signal which can...

After knowing the concepts of Digital signal processing, now it was time to apply those concepts on the dsp kit. In this experiment, we were given a general overview of programming on a dsp kit in assembly level language. The DSP kit used was TMS320f28375. Basic arithmetical​,logical and shifting Operations were performed on the kit. The instructions for above operations were implemented and register values before and after the execution were observed and noted down.

The FIR filter which is practicable realisable. The frequency response of the filter is multiplied with an appropriate function(known as window function) t o design a desired FIR filter  . On multiplication, we actually truncate infinite sample values of transfer function. Window function is selected on the basis of transition width criteria(which is filter width>= window width). We will obtain the desired FIR filter  b y the implementing appropriate window function with FIR filter algorithm.

In a real time system we receive signal values continuously. Thus when the signal is continuous, OAM/OSM computational methods are use to generate the reaponse of the system. This methods stores a finite amount of sample values from the input signal length of which is given by user, process it and generate the response with altering the values according to their respective algorithms​. The time required for this two technique is same.

Here we have designed our own Impulse invariant response (IIR) filter. First wh have designed Butterworth Low Pass filter and High Pass filter and plot the magnitude spectrum using scilab. Scilab was used as it is a open source software. Thus anyone can easily implement their mathematical model on scilab and analyse it. The magnitude spectrum  whicha was plotted after the input parameters provided by the us is in terms of Logarithmic scale on magnitude axis(Y-axis) and Frequency in Hertz(X-axis). To ensure correct nature of the magnitude spectrum we have taken the values in a certain range.

As the input signal values increases the DFT computational time also increases. Thus to decrease this time, we have implemented DITFFT radix-2 algorithm for N=4 and N=8. This algorithm decimates the input sample values into smaller pieces until the prime factor 2 is obtained.