Scientific Computing

Final verification of Intel vPro on site setup

When setting up a distant observatory it’s good to do a few final checks that otherwise require a site visit to fix. Although remote connection via SSH to the observatory PC is standard, the Intel vPro chipset, specifically the Intel Manageability Engine, serves as a crucial backup in case of operating system failure. This technology enables remote initiation of a Clonezilla USB stick or CD left in the PC to reimage the complete drive using the on-site external HDD prepared with Clonezilla prior to shipping. If this system fails, a site visit will be necessary to address major software issues.

To verify the Intel ME web server from the remote PC, access:

  • http://localhost:16992
  • https://localhost:16993

Verification from a laptop requires port forwarding SSH on the final hop, ensuring these ports are not exposed to the public Internet.

Access to the web interface allows selection of the boot drive. While this alone does not suffice for reimaging, maintaining two installed operating systems with boot-flagged partitions on each drive can be beneficial. Connection verification can be performed using MeshCommander, providing visibility and access to BIOS during boot and graphical remote control as if physically present at the PC.

Be sure the CPU fan hasn’t become loose from the CPU, that no wires are near the fan blades, and every cable is still plugged snug into the motherboard.

Using RFspace SDR-IQ on Linux with CuteSDR

Using the RFspace SDR-IQ on Linux is done with the CuteSDR program.

  1. install CuteSDR 
    apt install cutesdr
  2. In CuteSDR, click Setup → Network and enter 127.0.0.1 IP address.
  3. click Find → Find SDR then click OK. You should see the “run” button is available–even if you didn’t see a radio pop up under Find SDR.

Run

SDR-IQ on Linux is run using two terminals. Forking siqs_ftdi & didn’t seem to work for me.

Terminal #1:

siqs_ftdi

Terminal #2:

CuteSdr

Alternative

RFSpace SDR-IQ via SpectraVue on Linux WINE

Open source HF aeronomy radar project

A team of four ECE seniors has been self-selected and chosen to design and build an open-source aeronomy radar. The team will build up prior work by Juha Vierinen and David Hysell in the area of HF radar used to investigate the ionospheres behavior vs. time.

Their most recent published work involved one transmitter and two receivers. From discussion with Juha, these measurements become far more interesting when numerous measurement nodes are involved in a small geographic area extending perhaps 400x400 km with stations spaced every 50 km or so. This setup give the capability of telecentric “imaging” of the ionosphere at HF. Using offset antennas and dual synchronized receivers, polarization can be detected.

OpenCV CalcOpticalFlowHS Horn-Schunck

The default lambda=1.0 smoothness parameter for OpenCV 2.4 CalcOpticalFlowHS does not work well for some cases. Also, OpenCV’s lambda is different from Matlab Computer Vision Toolbox Horn Schunck opticalFlowHS('Smoothness',1.0) parameter, which also defaults to 1.0.

Horn Schunck Optical Flow comparison between Matlab and Python OpticalFlow_Python_vs_Matlab.py shows that OpenCV CalcOpticalFlowHS setting lambda=0.001 gives results that are much more like Matlab. OpenCV no longer has the original Horn Schunck optical flow.

Linearized Horn Schunck has problems with large displacements in the image, which can be resolved by a pyramidal algorithm.

Use SVG vector graphics in LaTeX

LaTeX can use “.svg” SVG vector graphics files directly without conversion via LaTeX svg package. If the LaTeX “.sty” doesn’t work with svg package, convert SVG to EPS instead.

This minimal working example of using .svg directly in LaTeX assumes file myfig.svg.

\documentclass{article}
\usepackage[clean]{svg}

\begin{document}

\begin{figure}
    \centering
    \includesvg[width=0.6\columnwidth](myfig.svg)
\end{figure}

\end{document}

Abandoned objects video databases

Events over the weekend in the US were mitigated by “see something, say something”-mindedness of passersby. Even better prompt detection of unusual events by ever more pervasive surveillance video may be achieved through coupling of machine vision-machine learning algorithms. Having a reference dataset is a useful starting point. Here are a quick list of abandoned object datasets.

11 videos of busy, day/night scenes

carefully collected few dozen videos presented in research paper

Mount .vdi VirtualBox disk image

Assume a VirtualBox guest OS Windows VDI disk image mydisk.vdi that you wish to mount without Virtualbox, perhaps because your guest OS crashed. The disk space used by the cloned .img file will be the full expanded size of the dynamic partition if one was used. Example: dynamic partition max size was 100 GB but only 10 GB was used → clone operation uses full 100 GB.

Rescue files from VDI by copying the VDI to IMG format:

VBoxManage clonehd --format RAW mydisk.vdi mydisk.img

cd ~/"Virtualbox VMs/Windows"

Examine partition

parted mydisk.img  #not sudo

unit
B
print

look for the “start” column for the biggest partition, that’s probably the virtual machine.

If you chose “12345 B”:

mount -t ntfs -o loop,ro,offset=12345 mydisk.img /mnt

It takes several minutes to unpack.

Explore the files in /mnt as usual. Copy out what you need.