Aims. We develop an automatic bubble-recognition routine based on Minkowski functionals (MF) and tensors (MT) to detect bubble-like interstellar structures in optical emission line images.
Methods. Minkowski functionals and MT are powerful mathematical tools for parameterizing the shapes of bodies. Using the papaya2-library, we created maps of the desired MF or MT of structures at a given window size. We used maps of the irreducible MT ψ2, which is sensitive to elongation, to find filamentary regions in Hα, [S II], and [O III] images of the Magellanic Cloud Emission Line Survey. Using the phase of ψ2, we were able to draw lines perpendicular to each filament and thus obtain line-density maps. This allowed us to find the center of a bubble-like structure and to detect structures at different window sizes.
Results. The detected bubbles in all bands are spatially correlated to the distribution of massive stars, showing that we indeed detect interstellar bubbles without large spatial bias. Eighteen out of 59 supernova remnants in the Large Magellanic Cloud (LMC) and 13 out of 20 superbubbles are detected in at least one wavelength. The lack of detection is mostly due to surrounding emission that disturbs the detection, a too small size, or the lack of a (circular) counterpart in our emission line images. In line-density maps at larger scales, maxima can be found in regions with high star formation in the past, often inside supergiant shells (SGS). In SGS LMC 2, there is a maximum west of the shell where a collision of large gas clouds is thought to have occurred. In the Small Magellanic Cloud (SMC), bubble detection is impaired by the more complex projected structure of the galaxy. Line maps at large scales show large filaments in the SMC in a north-south direction, especially in the [S II] image. The origin of these filaments is unknown.