Improve initial peak finding
Currently Atomap is optimized for analysing high quality STEM images with the same or similar structure over the whole image (for example https://ascimaging.springeropen.com/articles/10.1186/s40679-017-0042-5#Fig6).
However, many atomic resolution STEM images does not have the same structure across the whole image, like nanoparticles, aluminium precipitates or grain boundaries.
Currently the only possible parameter for finding the initial peaks is the minimum feature distance in the peak_local_max
function (http://scikit-image.org/docs/dev/api/skimage.feature.html#skimage.feature.peak_local_max), however this function has many different arguments which could be used to find the initial peak positions.
An easy improvement could be to be able to specify all the arguments in peak_local_max
, where an optimal value could be found by getting a multidimensional HyperSpy signal, where the navigation axes could be used to navigate across the parameter space.