From fowler at ece.msstate.edu Fri Dec 2 12:31:23 2011 From: fowler at ece.msstate.edu (James E. Fowler) Date: Fri Dec 2 12:31:39 2011 Subject: [ece8453] Re: About runlength code in Pr.6 In-Reply-To: <12CE41C5772847D987D30EC3B5015420@MyEdge> References: <4EC5234F.4080704@ece.msstate.edu> <12CE41C5772847D987D30EC3B5015420@MyEdge> Message-ID: <4ED9197B.9050109@ece.msstate.edu> > I have a Q about Pr.6 concerning implementing raster(Z). > When we do runlength coding along B0 -> H0 ->V0 ->D0->H1..., do we have > to start over runlength coding at a new subband or should we go through > without startover thinking virtually a big 1D sequence? > 1. B0 -> H0 ->V0 ->D0->H1 > separate rl code from B0 + separate rl code from H0 + ... > > 2. B0 -> H0 ->V0 ->D0->H1 > Consder this 2D into a 1D, then do rl code once. > Option 2 is more reasonable to me, but I want to make sure. > Deok. You want to do just a single runlength coding. This way, you can exploit runs of zeros from one subband to the next, which will likely occur frequently for the high-frequency subbands. -- ======================================================================= | Dr. James E. Fowler | | Billie J. Ball Professor and Graduate Program Director | | Department of Electrical & Computer Engineering | | Geosystems Research Institute (GRI) | | Mississippi State University | | | | fowler@ece.msstate.edu | | http://www.ece.msstate.edu/~fowler | | | ---------- Postal Address ---------------------- Contact Info --------- | Box 9571, 406 Hardy Rd. | ECE: Rm. 225, +1 (662)325-3640 | | Mississippi State, MS 39762 | HPC2: Rm. 208, +1 (662)325-2578 | | USA | Fax: +1 (662)325-2298 | ======================================================================= From fowler at ece.msstate.edu Mon Dec 5 11:40:04 2011 From: fowler at ece.msstate.edu (James E. Fowler) Date: Mon Dec 5 11:40:24 2011 Subject: [ece8453] Re: questions on the final exam In-Reply-To: References: Message-ID: <4EDD01F4.7030409@ece.msstate.edu> > A couple of clarifying questions: > Part 2 says "To implement each filter bank, use the m-file > analysis_filter.m that you wrote for Homework #4". We only wrote > circular_conv.m which later was replaced by mex-files to speed up > execution, the rest was downloaded from the class webpage. Can I just > use what's available from the webpage in this case or do I need to write > my own at this time? You can use any of the code we have developed over the course of the semester. > Part 6: > To do an inverse raster scan, if the image is not square, I need to pass > the row or column information as one of the arguments from encoder2 to > decoder2. Alternatively, I can include it at the beginning of z2 and > extract it out by the decoder. One additional number should not change > the entropy. But I can avoid all that if I assume the image is always > square. Can I make that assumption ? No, you cannot assume that the image is square. Put image-size information at the start of one of the sequences. It won't affect the bitrate substantially. -- ======================================================================= | Dr. James E. Fowler | | Billie J. Ball Professor and Graduate Program Director | | Department of Electrical & Computer Engineering | | Geosystems Research Institute (GRI) | | Mississippi State University | | | | fowler@ece.msstate.edu | | http://www.ece.msstate.edu/~fowler | | | ---------- Postal Address ---------------------- Contact Info --------- | Box 9571, 406 Hardy Rd. | ECE: Rm. 225, +1 (662)325-3640 | | Mississippi State, MS 39762 | HPC2: Rm. 208, +1 (662)325-2578 | | USA | Fax: +1 (662)325-2298 | ======================================================================= From fowler at ece.msstate.edu Mon Dec 12 10:48:19 2011 From: fowler at ece.msstate.edu (James E. Fowler) Date: Mon Dec 12 10:48:38 2011 Subject: [ece8453] Grades Posted Message-ID: <4EE63053.7090601@ece.msstate.edu> I have finished grading the final exams, and have posted the resulting statistics to the class website. Additionally, I have posted final course grades to Banner. I will be keeping the final exams. If you want to know your score, you can email me a request, and I will email your score to you. Alternatively, if you wish to see your exam, email me for an appointment. As a final note, I would discourage you from using the code that you have written this semester in your later research endeavors. The code you have written is very simplistic and doesn't implement a variety of functionality useful in real-world applications (biorthogonal wavelets, symmetric extension, arbitrary-length input signals). Additionally, what you've implemented is very slow, and may not have been thoroughly tested. Many people use the Matlab Wavelet Toolbox toolbox, but I have found it cumbersome. I recommend instead Ivan Selesnick's package (http://eeweb.poly.edu/iselesni/WaveletSoftware/); alternatively, if you're working in C/C++, use QccPack (http://qccpack.sourceforge.net). Both of these packages offer much greater functionality than we have implemented this semester. Have a great break! -- ======================================================================= | Dr. James E. Fowler | | Billie J. Ball Professor and Graduate Program Director | | Department of Electrical & Computer Engineering | | Geosystems Research Institute (GRI) | | Mississippi State University | | | | fowler@ece.msstate.edu | | http://www.ece.msstate.edu/~fowler | | | ---------- Postal Address ---------------------- Contact Info --------- | Box 9571, 406 Hardy Rd. | ECE: Rm. 225, +1 (662)325-3640 | | Mississippi State, MS 39762 | HPC2: Rm. 208, +1 (662)325-2578 | | USA | Fax: +1 (662)325-2298 | =======================================================================