- contact -
logo

Documentation

Random point counts

Description

Random point counts are commonly used in ecological sampling to statistically estimate percentage coverage of organisms and associated substrate from image samples. The idea is that instead of exhaustively counting each and every organism, one could superimpose on the image a number of randomly distributed points, visually identify and count the organisms beneath each point, and use these counts as a proxy of the actual population characteristics.

Highlights

  • Random points operate on a separate layer

  • Custom, user-defined number of random points

  • Three different spawn modes: random, stratified random, uniform

  • Random points can be optionally confined within the quadrat's active area

  • Multiple points can be simultaneously selected, assigned to species, re-assigned or reset

  • Output and descriptive statistics graphs available with a single click

  • Percentage coverage is exported both at species level and total quadrat level (grouping all random points irrespective of species); statistics for unassigned points are also exported



Read more: | User interface | Customize the spawn canvas | Customize the distribution | Select points |

| Assign to species | Point properties | Descriptors and worksheet output | FAQs |



rndpnts

User interface

Random points are controlled through a dedicated panel in photoQuad's interface (Markers/Points tab):

(A) Number of points and spawn method

The mutually exclusive checkboxes control the points' spawn method, while the "N points" editbox controls the total number of points spawned.

(B) Options

Use this panel to control the random points' spawn canvas, the visibility and color of the stratification grid, and to suppress the confirmation message displayed when a new set of random points is spawned while random points already exist on the image.

(C) Appearance

Customize the appearance of random points; you can change these attributes anytime during processing.

(D) Launch buttons

Click the "Spawn" button to spawn points using the current settings. The lasso selection tool can also be launched from here.

rndgui


Customize the spawn canvas

The spawn canvas is the region in which random points are allowed to spawn. This canvas can either be the whole image, or the quadrat's effective sampling area:

1) loose canvas: random points are allowed to spawn anywhere on the image, irrespective of whether a quadrat has been detected or not

rndloose

2) crop to quadrat: random points are automatically confined within the quadrat, if a quadrat has been detected

rndcrop

The spawn canvas (i.e. the quardat) does not have to be rectangular; any arbitrary polygon would do:

rndcrop


Customize the distribution

Not only the random points' spawn canvas can be customized, but also their distribution, i.e. the spatial pattern according to which points are superimposed on the image. photoQuad supports three distribution modes: random, stratified random, and uniform.

Random

Each image pixel has the same probability of spawing a point. This is as random as it gets, however clustering may occur in particular regions, while some other regions may get under-sampled.

thumb

Stratified random

The spawn canvas is divided into sub-cells, and points spawn within each cell in a random manner. The method ensures that at least one point is present in each cell, thereby reducing the clustering and under-sampling issues of the previous method.

thumbrandom

Uniform

The spawn canvas is divided into sub-cells, and points spawn right at the center of each cell. The result is a uniform distribution of cells throughout the canvas.

thumb


Select points

In order to assign one or multiple random points to a particular species library entry, photoQuad needs to know which points you refer to. Random points have two possible modes regarding their selection status:

  • sleep: ..zzZ

  • selected: this mode refers to the activated points, i.e. those that are selected in order to be assigned to a species. Left-click a selected point to deselect it and switch it back to "sleep" mode.

The following images illustrate the point selection process. Note that when selecting a point with the mouse, the interactive object is the actual point marker, not its text label:

1) start

poinselection

2) hover to highlight

poinselection

3) click to select a point..

poinselection

..or use the lasso for multiple

poinselection

(press "s" to preview)

poinselection

4) DONE: three points selected

poinselection

5) right-click any point to assign the whole group

poinselection

Tip: the lasso tool and its keyboard/mouse shortcuts work exactly like the freehand region drawing tool.



Assign to species

As described in the previous section, selected points can be assigned to a species, so that the software can calculate species coverage and related descriptors. The following image shows the "Assign random points" user interface; please hover the mouse over the image to see the appropriate description.

When reading through this example, consider that yellow points are unassigned, while green ones are points that have already been assigned to a species in a previous step. A random mixture of them is currently selected for demonstration purposes.

hover mouse over the dotted rectangles to see description


This is not easy to demonstrate with a static image like the one used above, however modifications either on the "Assign random points" window or on the actual points on the image are linked in real-time. This means that selecting a point with the mouse (or the lasso) automatically brings it to the appropriate "Select" or "Sleep" listbox, and vise-versa. Move around points across listboxes, and their status is updated on the image accordingly. Click a point in a listbox, and its location is highlighted on the image so that you don't have to look for it.



Querying random point properties

To query a point's properties, right-click it and select "Current point properties". The properties window lists the point's identification number, its species association, its current selection status, as well as a list of all other points that are assigned to the same species. Click the toggle button with the light bulb icon to isolate those points on the image; its a quick way to review your work progress.

thumb


Descriptors and worksheet output

The random points statistics panel is an extension of the "Assign random points" window. It provides information regarding the number of species, their species name and ID, the number of points assigned per species, as well as the equivalent percentage against the total number of species. For each species category there is an option for allowing the points of that category to remain visible or hide. This way points of any given species can be displayed separately on the image, therefore enhancing image processing and making reviewing easier.

Unassign one or multiple random points

Click a species category in the summary table and the points that are currently assigned to that species are displayed in the small panel below the table; with respect to this demonstration, "Hexadella racovitzai" is selected. Use the buttons next to the listbox to reset a point's species association status.

the random points statistics panel

thumb

Click the "Summary & export options" button, or right-click a point and select "Show summary" to navigate to the worksheet export window:

thumb

For each species category (including "unassigned" points), a line with the following descriptors is exported, comma separated (*.csv)

Descriptor Units Definition
Image text string Image filename
Path text string Image directory path
spp Name text string Species name
spp ID - Species ID (-1 for unassigned points)
N pts per species - Number of points assigned to this species
Cov% per species % Percentage coverage of this species (max 100%)
N pts ALL species - Total number of points assigned, summing all points irrespective of species
Cov% ALL species % Total percentage coverage of points assigned, summing all points irrespective of species (max 100%)
Reference random pts - Number of random points spawned

..read more on worksheet output



Important !

When exporting your data, make sure that the target file is not simultaneously opened in Excel or in another application.

Excel takes hold of the file and prevents other applications from having write-access to it, so photoQuad will not be able to export the requested data. If this happens, you will receive a message saying so. Please close the target file from Excel and repeat the export.




Frequently Asked Questions


Are random points automatically confined into the quadrat's active area?

Yes, but you have the option to disable this feature.

Do random points depend on image calibration?

No. Random points carry no distance or area information, and therefore don't depend on calibration data.

Can I re-assign one or multiple random points during analysis?

Yes. Select them, right-click, select "Assign to species" from the context menu, and choose the appropriate library entry from this interface.

Can I customize the color of assigned points, so that it differs among species?

No. In the current implementation, two categories are only separately colored: "unassigned" and "assigned" points. You cannot have separate coloring within these categories. We'll fix that in an upcoming version.

Can I reset a random point i.e. switch it back to unassigned?

Yes. Please read this section.

What happens if I try to assign some random points, but I have not yet imported a species library?

photoQuad will automatically prompt for a custom species library, or suggest to use the embedded one.

Can I toggle random points' visibility on/off?

Yes. Use the corresponding checkbox in photoQuad's floating GUI.

When I use the "Stratified random" or "Uniform" distribution, I sometimes see more than one point in each stratification sub-cell. Why?

This is how the stratification algorithm works. If you ask for N points total to spawn, the software tries to stratify the spawn canvas in a grid of n rows by n columns, i.e. n2 sub-cells. To do that, it looks for an n value that is the largest square root that is equal to or smaller than N. For example, if N=100, cells will be 10-by-10 and each cell is populated with 1 point. if N=36, cells will be 6-by-6 and each cell is populated with 1 point. But what if N=63? Cells will be 7-by-7 (=49), each cell will be populated with 1 point, and the remaining 14 points are randomly spawned anywhere on the image; total is 63, but some stratification cells will have more than one point within them.