{
  "id": 38764,
  "title": "ahi to image thread",
  "url": "/competitions/passenger-screening-algorithm-challenge/discussion/38764",
  "author_name": "",
  "post_date": "2017-08-30T15:04:41.506161200Z",
  "votes": 4,
  "comment_count": 1,
  "views": 0,
  "content": "<p>It seems like this paper explains some method for reconstructing images from raw data domain, I don't know if the data pre-calibration performed on the tsa dataset actually makes it hard to implement this. I'm still trying to get up to speed with the domain knowledge.</p>\n\n<p><a href=\"https://www.researchgate.net/publication/318976373_Three-Dimensional_Microwave_Imaging_for_Concealed_Weapon_Detection_Using_Range_Stacking_Technique\">https://www.researchgate.net/publication/318976373_Three-Dimensional_Microwave_Imaging_for_Concealed_Weapon_Detection_Using_Range_Stacking_Technique</a></p>",
  "messages": [
    {
      "id": "217420",
      "postDate": "08/30/2017 15:04:41",
      "content": "<p>It seems like this paper explains some method for reconstructing images from raw data domain, I don't know if the data pre-calibration performed on the tsa dataset actually makes it hard to implement this. I'm still trying to get up to speed with the domain knowledge.</p>\n\n<p><a href=\"https://www.researchgate.net/publication/318976373_Three-Dimensional_Microwave_Imaging_for_Concealed_Weapon_Detection_Using_Range_Stacking_Technique\">https://www.researchgate.net/publication/318976373_Three-Dimensional_Microwave_Imaging_for_Concealed_Weapon_Detection_Using_Range_Stacking_Technique</a></p>",
      "rawMarkdown": "It seems like this paper explains some method for reconstructing images from raw data domain, I don't know if the data pre-calibration performed on the tsa dataset actually makes it hard to implement this. I'm still trying to get up to speed with the domain knowledge.\n\nhttps://www.researchgate.net/publication/318976373_Three-Dimensional_Microwave_Imaging_for_Concealed_Weapon_Detection_Using_Range_Stacking_Technique",
      "votes": null
    },
    {
      "id": "223795",
      "postDate": "09/23/2017 16:20:01",
      "content": "<p>that's pretty hardcore!\nI wonder if it's worth pursuing?</p>\n\n<p>I'm speculating that in terms of quality, you probably can't do much better than the processing already offered in the other 3 formats, but perhaps there is something to be gained in terms of resolution? In any case it's probably mostly fft/ifft and computationally discrete, and therefore there is a size parameter which could be tweaked. The 512x512 smells a bit like discrete fft :) And since this is supposedly a millimeter technology, there might be something to be gained from actually doing it.. personally I've never done a fft on a huge chunk like this and my old Matlab on my old 2gb Mac PPC might suffer :) But my main goal in this competition is to learn to do it in the cloud in Python.</p>",
      "rawMarkdown": "that's pretty hardcore!\nI wonder if it's worth pursuing?\n\nI'm speculating that in terms of quality, you probably can't do much better than the processing already offered in the other 3 formats, but perhaps there is something to be gained in terms of resolution? In any case it's probably mostly fft/ifft and computationally discrete, and therefore there is a size parameter which could be tweaked. The 512x512 smells a bit like discrete fft :) And since this is supposedly a millimeter technology, there might be something to be gained from actually doing it.. personally I've never done a fft on a huge chunk like this and my old Matlab on my old 2gb Mac PPC might suffer :) But my main goal in this competition is to learn to do it in the cloud in Python.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 223795,
      "author_name": "tothink",
      "author_url": "",
      "post_date": "09/23/2017 16:20:01",
      "content": "<p>that's pretty hardcore!\nI wonder if it's worth pursuing?</p>\n\n<p>I'm speculating that in terms of quality, you probably can't do much better than the processing already offered in the other 3 formats, but perhaps there is something to be gained in terms of resolution? In any case it's probably mostly fft/ifft and computationally discrete, and therefore there is a size parameter which could be tweaked. The 512x512 smells a bit like discrete fft :) And since this is supposedly a millimeter technology, there might be something to be gained from actually doing it.. personally I've never done a fft on a huge chunk like this and my old Matlab on my old 2gb Mac PPC might suffer :) But my main goal in this competition is to learn to do it in the cloud in Python.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "217420": "It seems like this paper explains some method for reconstructing images from raw data domain, I don't know if the data pre-calibration performed on the tsa dataset actually makes it hard to implement this. I'm still trying to get up to speed with the domain knowledge.\n\nhttps://www.researchgate.net/publication/318976373_Three-Dimensional_Microwave_Imaging_for_Concealed_Weapon_Detection_Using_Range_Stacking_Technique",
    "223795": "that's pretty hardcore!\nI wonder if it's worth pursuing?\n\nI'm speculating that in terms of quality, you probably can't do much better than the processing already offered in the other 3 formats, but perhaps there is something to be gained in terms of resolution? In any case it's probably mostly fft/ifft and computationally discrete, and therefore there is a size parameter which could be tweaked. The 512x512 smells a bit like discrete fft :) And since this is supposedly a millimeter technology, there might be something to be gained from actually doing it.. personally I've never done a fft on a huge chunk like this and my old Matlab on my old 2gb Mac PPC might suffer :) But my main goal in this competition is to learn to do it in the cloud in Python."
  },
  "source": "meta"
}