Metadata Extraction#2
Conversation
|
|
||
| %% Crop metadataRef with Same ROI | ||
| metadata = imread([refPath '\metadataRef.png']); | ||
| metadata=imcrop(metadata,[115 185 1900-115 540-185]); |
There was a problem hiding this comment.
I don't see where this is being used anymore. You should add something like this after the metadata imcrop.
metadata = imbinarize(rgb2gray(metadata));
img(~metadata) = 255; %% making everything other than the omr points of interest white
img = ~imbinarize(rgb2gray(img)); %% now you have only the omr blobs as white in the img
| medium= s{4,2}; | ||
| division= s{5,2}; | ||
|
|
||
| age_crop=imcrop(img,[age(1)+50 age(2) 105 45]); |
There was a problem hiding this comment.
The output of this function should be something like this:
age = '18-25'
gender = 'Other'
dom_hand = 'Left'
not image crops. This metadata will be directly appended to a csv file in the pipeline.
| @@ -0,0 +1,72 @@ | |||
| function [age_crop, gender_crop, dominant_hand_crop, medium_crop, division_crop]= metadataExtraction(scanned_img, refPath) | |||
There was a problem hiding this comment.
the function should take the registered image and metadata reference image as input, and output exact metadata information (gender = 'other'), not image crops.
There was a problem hiding this comment.
This function will be placed inside extract.m right after surfAlignGPU is performed. So you'll already have the registered image by then.
No description provided.