measures2.0_decomposeOve

[xulman]

Recently a couple of ideas pilled up of what everything else could be measured. Here in particular:

Ove has seen a correlation with SEG (average) performance... which is strange as Ove was

• designed mainly (in a hope) to reflect TRA performance
• should measure how much a cell is translating

But in reality, Ove is also influenced by (stationary) shape changes like cell growth, "blobbing", rotating (if elongated)...
Maybe rather than decomposing shape changes, let's attempt to "subtract" cell movement itself.

Counterexample of low Ove that should give good SEG: Consider a cell that is not changing is texture, shape etc... just appears on different close-by position in the image: It is the same pattern (if segmentable once, should be segmentable always) with low Ove.