Global feature means looking the whole image while local means focusing on something. In a group photograph, you may look all 10 faces in the group and may not recognize some of them. You may only on one face then this become local. In a face, nose, eyes etc are local features. Wavelets are capable to give you local as well as global features.  

i agree with Mittal,

Local-- more relevant information like details about individual part in a tree in a mountain

Global- overall information- view of trees available in a mountain

Global: Same rule applyied to all pixels; local: Rule only applied to select=ed regions of interest (such as a 3x3 matrix).

The only criteria is that whether the feature vector represents the entire image or some parts of the image. If it represents the entire image, then it should be a global feature, otherwise it should be a local feature.

how regional features are different from global?

can local ,global,regional features can be extracted by means of color or texture or shape ?

Global feature means viewing the whole image while local means concentrating on something.