Leginon

h1. Matlab Target Finder

The Matlabe Target Finder node behaves like other target finding nodes in that it takes

images as input and outputs a list of targets. In this case, Leginon sends the image to an

externally running Matlab process, which then sends the list of targets back to

Leginon.

In order to use the Matlab Target Finder, you must:

<highlights>

<simplelist type="vert">

<member> have matlab installed on the machine where the node will run</member>

<member> have the pymat python module installed (pymat.sourceforge.net)</member>

<member> include the path of libeng.so in the environment variable LD_LIBRARY_PATH. This

library is dynamically linked by pymat in order to connect with matlab.</member>

</simplelist>

</highlights>

In addition to the usual target finder settings, this node needs to be configured with the

matlab function (.m file) that should be executed on the image. The function should take the

image as input in the form of a two dimensional array. The output should be two arrays, one

for focus targets and one for acquisition targets. Each is an array of x,y pairs. Here is a

simple example where the function generates a constant set of targets for any input image:

<pre>% example Matlab module for target finding

function [acquisition, focus] = example(image)

acquisition = [168, 234; 203, 54; 450, 390]

focus = [324, 42; 36, 248; 151, 94]</pre>

This generates three acquisition targets and three focus targets. Leginon will display

the targets on the image and optionally allow the user to edit the targets before submitting

them.

Required bindings for recieving images and publishing targets:

<highlights>

<simplelist type="vert">

<member> previous Acquisition - (AcquisitionImagePublishEvent) -> Hole Finder</member>

<member> Hole Finder - (ImageTargetListPublishEvent) -> next Acquisition</member>

<member> Hole Finder - (QueuePublishEvent) -> next Acquisition</member>

</simplelist>

</highlights>

Required bindings for proper waiting among nodes:

<highlights>

<simplelist type="vert">

<member> Hole Finder- (ImageProcessDoneEvent) -> previous Acquisition</member>

<member> next Acquisition - (TargetListDoneEvent) -> Hole Finder</member>

</simplelist>

</highlights>

h2.  Settings

*  Allow for user verification of picked holes = When enabled, this feature will

pause and wait for the user to verify whether the correct exposure and focus targets

have been selected after the Hole Finder finished processing the input image.

*  Queue up targets = When enabled, submitting targets do not publish them

immediately but to put them in a target list queue.

*  Declare drift when queue submitted= When enabled, the node forces target shift

correction when the queue is submitted. Useful when accurate targeting is

needed.

*  Matlab Module File = the entry defines the Matlab script file used in the target

finder.

h2.  Image Control Panel

NOTE: To see the effects of Testing the settings for a particular Hole Finder step in

the Image Control Panel, the corresponding Hole Finder step Display must be enabled before

using the Test feature in that step's Display Settings.

h2.  Image

h2.  Display

To view the original input image, enable this button.

h2.  Display Settings

*  Test Image: File Entry = Enter the path and name of a MRC image that the

Matlab Target Finder can be tested on.

h2.  acquisition

In Matlab target finder, the purpose of the acquisition setting is only for display

and user selection.

h2.  Display

To view acquisition, the Image Display must be enabled. The acquisition targets

that the finder chooses will be shown with a green crosshair.

h2.  focus

In Matlab target finder, the purpose of the focus setting is only for display and

user selection.

h2.  Display

To view focus, , the Image Display must be enabled. The focus targets that the

finder chooses will be shown with a blue crosshair.