Out of convolution operation and inverse filtering operation,which operation is better regarding the stability point of view(minimum...

Out of convolution operation and inverse filtering operation,which operation is better regarding the stability point of view(minimum phase system), if i introduce noise into system. Thanks Riz

I am the actuary for a reinsurance company in Bermuda. I have come across a problem that would be solved faster using the FFT. all i need to use...

I am the actuary for a reinsurance company in Bermuda. I have come across a problem that would be solved faster using the FFT. all i need to use it for is convolution. For instance does anyone have a code that would convolute (.26274,.21074,.15547,.11472)*(.26274,.21074,.15547,.11472) using the FFT? I would actually be convoluting 2^13 or 8192 numbers in each vector instead of 4. I just want to ge...

Hi - I'm interested in computing 2d convolutions of matrix X with arbitrary kernel K and its non-conjugate transpose K'. Is there a way to...

Hi - I'm interested in computing 2d convolutions of matrix X with arbitrary kernel K and its non-conjugate transpose K'. Is there a way to compute conv2(X,K) and conv2(X,K') faster than doing 2 generic convolutions? Also, if kernel KK has radial symmetry, can conv2(X,KK) be computed faster than doing a generic convolution? Thanks!

Hi folks, I was experimenting with convolution, using the the overlap-add ideas with FFT. I was using an impulse of fading white noise to test...

Hi folks, I was experimenting with convolution, using the the overlap-add ideas with FFT. I was using an impulse of fading white noise to test the algorithm for smoothness, and I noticed to made an almost perfect diffusion sound. My question; is it possible to feed the FFT with white noise and a mixture of the previous noise or something to create a nice diffusion fall-off without us...

Hi all, I intuitively had an idea that f(t) ** g(t) = f'(t) ** int(g(t)) = int(f(t)) ** g'(t) where " ' " denotes differentiation,...

Hi all, I intuitively had an idea that f(t) ** g(t) = f'(t) ** int(g(t)) = int(f(t)) ** g'(t) where " ' " denotes differentiation, int(f(t)) denotes integration, " ** " denotes convolution. I want to prove it is true but I met with some difficulties for which I need your help! Let me first define convolution as integral from -infinity to +infinity, if I define int(f(t)) to be ...

Hello group, It is my pleasure to be posting queries to this group again. My understanding of the presence of the cyclic prefix (CP) in an...

Hello group, It is my pleasure to be posting queries to this group again. My understanding of the presence of the cyclic prefix (CP) in an OFDM symbol is to: 1. Prevent inter-symbol interference in a FIR channel if the length of the CP is larger than the delay spread of channel. 2. Makes linear convolution "appear" as circular convolution over the duration of the useful period in the OF...

The derivative of a 1D signal can be obtained from convolution with the kernel [1 -1]. To get the derivative of a 2D image, can someone help me...

The derivative of a 1D signal can be obtained from convolution with the kernel [1 -1]. To get the derivative of a 2D image, can someone help me with what the kernel is - something like [1 -1] [-1 1]? I guess. And for the second derivative, in 1D it's [-1 2 -1] and for 2D [ 1 -2 1] [-2 4 -2] [ 1 -2 1]? Thanks for helping!

Hi I am trying to implement a linear convolution of FIR IR and a large input sequence. The IR is symmetrical (L=11). I receive the data in...

Hi I am trying to implement a linear convolution of FIR IR and a large input sequence. The IR is symmetrical (L=11). I receive the data in frames of say 128 samples (could be less or more than this). The output exhibits problems at the frame boundaries. What is the best way of overcoming these boundary effects. I have tried saving L-1 input samples from the end of the previous frame, which I ta...

Hello, I am thinking about a question about how to figure out whether a linear filter really filters out anything against a particular...

Hello, I am thinking about a question about how to figure out whether a linear filter really filters out anything against a particular sequence (or any sequence). Here I don't want to compute w/ convolution, but I don't know if it is possible to think in a convolution way for my problem. For example, I might have 2 filters h1 and h2,where h1 = [-1 0 1], h2 = [-1 2 -1] Their magnitude...

I got a simple question about performing a FFT convolution in the polar form, how do you do it? I expect it to be something like this : DC...

I got a simple question about performing a FFT convolution in the polar form, how do you do it? I expect it to be something like this : DC out = DC in * DC k (* meaning multiply) Mag out = Mag in * Mag k Phase out = Phase in + Mag k k being the kernel (no need to explain what in and out stand for i guess) I suck too much at maths to make sure it is correct, can anyone out there te...

Quick (and dumb) question, but just wanted to make sure I understand. Supposed I have an input vector x of length N=500. And I also have a...

Quick (and dumb) question, but just wanted to make sure I understand. Supposed I have an input vector x of length N=500. And I also have a filter vector h of length M=3000. If peforming convolution filtering via FFT, what should I set the FFT sizes to be? I want the output signal to be the same size as the input signal (N) I assume it would need to be greater than N+M-1 to not cause circular c...

Hi, i searched the web for quite some time now and just don't grok how to normalize when doing partitioned frequency domain...

Hi, i searched the web for quite some time now and just don't grok how to normalize when doing partitioned frequency domain convolution. Basically for an unnormalized FFT/IFFT pair (i use fftw, so all the FFT/IFFT i use is unnormalized), the necessary normalization factor would be 1/N applied once or 1/sqrt(N) applied twice. But now i have IFF(FFT(signal)*FFT(response)) and i wonder wh...

Hi, I'm using FFT convolution for smoothing. Basically FFT kernel is exp(-x^2) and the signal is something positive. All of it is done in a...

Hi, I'm using FFT convolution for smoothing. Basically FFT kernel is exp(-x^2) and the signal is something positive. All of it is done in a single pass, so no need for overlap-add and stuff like that. Seems working fine, except now I tried it on quite long data - the FFT length was 262144, kernel being 32768 samples (zero-padded to 262144), calculated in 32-bit floating points, FFT implemen...

TITLE:"Fast Convolution (FFT) Filtering: From Basics to Filter Banks". DESCRIPTION: Fast Convolution filtering is a powerful technique with...

TITLE:"Fast Convolution (FFT) Filtering: From Basics to Filter Banks". DESCRIPTION: Fast Convolution filtering is a powerful technique with which every DSP engineer should be familiar. All but the shortest FIR filters are more efficiently implemented with FFTs rather than direct forms. The longer the filter; the greater the advantage. This presentation begins with the primary forms o...

I'm looking to prove circular convolution using the Discrete Hartley Transform. I saw an old thread on this forum where someone had got it...

I'm looking to prove circular convolution using the Discrete Hartley Transform. I saw an old thread on this forum where someone had got it working: http://www.dsprelated.com/showmessage/130417/1.php I'm using the same formulations given by wiki: http://en.wikipedia.org/wiki/Discrete_Hartley_transform http://en.wikipedia.org/wiki/Hartley_transform (see the "properties" section in ea...

Hi, My problem is the following: Given a integer N, I need to generate a NxN kernel that approximates an arbitrary frequency filter. So I can...

Hi, My problem is the following: Given a integer N, I need to generate a NxN kernel that approximates an arbitrary frequency filter. So I can use the (small) kernel to filter the image in spatial domain with convolution operations. For example I use the lowpass filter: (lowLv-hiLv) * e^ [-(f/Ft)^2] + hiLv where f is the euclidean distance between center and a "pixel" in frequency domai...

My problem is not dsp related, but the method is closely related to methods in dsp, so I am hoping to get some help here. I am using the...

My problem is not dsp related, but the method is closely related to methods in dsp, so I am hoping to get some help here. I am using the symmetric cubic convolution kernel ("Catmull-Rom splines") to interpolate data over a limited range in a variable x. For the interpolation I am using typically 10 nodes which are equidistant in x. Example: The interpolated function between nodes 4 and 5 i...

Hi all, I figures out that the maximum number for which MATLAB can give an answer to factorial is 170. This triggered a thought in my mind I...

Hi all, I figures out that the maximum number for which MATLAB can give an answer to factorial is 170. This triggered a thought in my mind I want to share with you guys. What if I take number in character string form as input to factorial, extract the number sequence as number array use convolution to do multiplication and then given the answer back in character? In this way the max range can ...

In graphical convolution either the system impulse response h(n) or the input x(n) is folded/flipped and then slid across the other to...

In graphical convolution either the system impulse response h(n) or the input x(n) is folded/flipped and then slid across the other to determine the system response y(n). Why is this so? I have reasoned this as being necessary to align the "present" input to the "present" output of the system response to get the correct system time output. Is this reasoning valid? None of the several signals and s...

Hi all, I want to filter out very narrow band signal, and thinking about using fast-convolution filtering. The input signal is real (not...

Hi all, I want to filter out very narrow band signal, and thinking about using fast-convolution filtering. The input signal is real (not analytic) and the output signal should be real. I wonder if there is any trick I can play to take advantage of real signal processing? Is there any saving on computation I can achieve since I deal with real signal? I have huge data to process, thus, even sma...