# How can we know the geometry of a blob detected by OpenCV

I am performing blob analysis using connectedComponentsWithStats, which returns elementary statistics about the blobs (AABB and area). It is also possible to compute geometric moments.

Anyway, I need to compute my own features and for this I need the detailed geometry, be it a chain of outline pixels, a run-lengh-coded description or similar.

I know that the contours can be obtained by other means (findContours), but I would like to know if the connected components engine can help me achieve my goal or if it is a black box forever.