I am working on a machine learning project processing fundus images.
I have an image dataset, each image in which comes with a 5-number tuple called "ellipse parameters". However, the dataset is poorly documented and I can't figure out what the fifth parameter is.
Here are some examples of the five numbers:
img1:195,199,222,227,0
img2:168,171,209,214,-39
img3:323,344,409,420,14
I am pretty sure believe that the first four numbers are the coordinates of the two focus points (because their locations on the corresponding images make sense). But the fifth number is weird -- it doesn't look like either the sum of distance to the two focus points or the eccentricity.
My Question:
So, what could this number be? And how can I get the ellipse function ?
Update #1:
The dataset contains multiple (image,txtfile) pairs.
In the txtfile, there are some numbers as listed below:
I) a center point (x,y coordinates, 2 numbers)
II) 16 boundary points (x,y coordinates, 16*2 = 32 numbers, they can form an polygon)
III) "ellipse parameters" mentioned above (5 numbers)
In order to better understand how these numbers work, I did the following:
1) draw the center point (the blue point in the following images),
2) draw the polygon by connecting the 16 points (the green boundary),
3) draw a line segment between the two focus points (the green line, however, I am now not sure whether they are the two focus points)
4) print the 5th number on the left upper corner of that image.
Now, I have generated images like this:
img_285, and its corresponding data from its txt file:
I) center point: 255, 251
II) 16 boundary points: (420, 251), (413, 316), (380, 376), (323, 416), (255, 425), (191, 405), (140, 366), (106, 313), (90, 251), (97, 186), (130, 126), (187, 86), (255, 77), (319, 97), (370, 136), (404, 189)
III) "ellipse parameters": 255, 251, 322, 357, -29
I notice that the first focus point (maybe it will turn out to be something else, but at least for now I still call (255,251) and (322,357) the two focus points) is always the same as the center point, and that's why in the images you will always see the line segment starts from the blue point.
Here are some more images with their "ellipse parameters":
img_290 : 231, 239, 311, 290, 39
img_291 : 206, 196, 297, 314, -134
img_293 : 206, 182, 168, 211, -21
img_294 : 170, 158, 158, 242, 0
img_317 : 231, 255, 285, 260, -43
img_332 : 287, 288, 283, 323, 23
After checking these images, I am now even more confused. I feel like the (3rd number, 4th number) pair may not be a focus point.
Why would I think so? I don't want to talk too much about Ophthalmology. In short, if there is a way to interpret these "ellipse parameters" and draw an ellipse, that ellipse should locate within the polygon, or at least not exceed the boundary of the polygon too much.
However, in img_291, the "focus point" (3rd,4th), or (297, 314), is out of the boundary of the polygon (as the line segmentation exceeds the polygon). If I draw an ellipse with these two points((206, 196) and (297, 314)) as foci, that ellipse must have a pretty large part outside the polygon. This perhaps indicates that (297, 314) is not a focus point at all, thus there will be 3 parameters (instead of 1 ) that I need to understand.
Of course, img_291 could be an outlier that should be excluded. No matter which is the case ((3rd,4th) be a focus point or not), how should I interpret the 5 ellipse parameters?