Crates converting binary masks to geo_types::geometry::Polygon

help
1

I was looking to convert binary masks into geo_types::geometry::Polygon. Was wondering if there were crates which allowed this deserialisation?

More concretely, given an ndarray object:

[
        [0, 1, 1, 0],
        [1, 1, 1, 1],
        [1, 1, 1, 1],
        [0, 1, 1, 1],
        [0, 0, 1, 0]
    ]

I wanted to convert it to a polygon:

polygon![
    (x:4, y:2),
    (x:2, y:0),
    (x:0, y:2),
    (x:2, y:4),
    (x:4, y:2),
];

I could find crates that rasterise polygons into a binary mask such as geo-rasterize, but nothing going the other way. Am I missing anything?

2

This problem is generally known as finding contours. rten_imageproc::find_contours has an implementation of this based on the algorithm that OpenCV's findContours function uses. It uses a different tensor library (rten-tensor), but there is some code here which shows how to convert an ndarray view into an equivalent NdTensorView. Alternatively you could copy and modify the find_contours implementation.

The rten-imageproc crate also has functions to simplify polygons (that is, find a polygon for the mask with fewer vertices but not as tight a fit to he original mask) and get rotated bounding rectangles containing a polygon.

Edit: One assumption I'm making here is that you want to treat each group of connected pixels in the mask as a separate polygon. If the mask has gaps but you want to get a single polygon as output, you could merge the polygons by taking the convex_hull of all the polygons' points.

1 Like
3

This topic was automatically closed 90 days after the last reply. We invite you to open a new topic if you have further questions or comments.