Recipe for analysing masculinity from shape

By: Dan Re

Put all men and women’s face tems /jpegs in one directory.

Make sure all aligned to same grand average of men and women.

Reducing templates (Updated 20 December 2011)

Used for morphological masculinity analysis, etc.

This recipe is for deleting lines from templates, which is useful for conducting PCA on face shape. We used this to create morphological masculinity scores.

THIS RECIPE WORKS WITH THE NEWEST VERSION OF PSYCHOMORPH, WHICH HAS A PLUGIN CALLED “BATCH BATCH DELETE LINES.”

Step 1: Create a list of jpegs and templates for the images you want to delete lines from (the list would look the same as a list for averaging faces).

Step 2: Create a list of points that you want to delete. You can find the point labels by clicking “View>Draw Labels” in the Transform (single) window. Create the list by naming the points with a space in between. For example, to delete the left and right cheekbone lines in a 3dsk face, you would create the following list:
“164 165 166 167 168 169”.
Make sure to save the list of images and templates and the list of points in the same directory.
Step 3: To batch delete points, go to the Transform window and click “Plugins>Perception Lab.jar>Batch Batch Delete Points.” This will prompt you four times.
Step 4: Open the LIST OF POINTS you want to delete first.
Step 5: Open the LIST OF IMAGES AND TEMPLATES second.
Step 6: Name your output directory
Step 7: Enter a string to append
This should create a new set of templates without the lines for the points you deleted.

PCA

Build asm file

Select menu option (Dual window>PCA>Build Shape PCA (ASM)>Create file name for saving> Open list of trimmed_templates).
Create PCA of shape >n components
Select menu option (PCA>Batch PCA Analyse Shape> ASM file> Save output directory> List of images and tems).
This creates a list of PCA components of decreasing importance (variance explained (see bottom line) and the list has the same order of faces as the list of images and tems

Note: PCA components do not come out in terms of decreasing importance in Psychomorp

Using excel, determine which components in the model explain greater than average variance and truncate the model to include only these.

Stage 3 building shape model predicting sex

Take into spss for regression/discriminate analysis   (incomplete)