Digital Image Correlation (DIC) can be used as answered by Khaled to derive true stress-strain values. One such system is Aramis 3D 5M optical system which can record instantaneous 3D deformation and strain of tensile specimens during test. Please visit the following website for more details.

http://www.gom.com/metrology-systems/aramis.html

You can calculate etrue=ln(1+eeng) and strue= seng(1+eeng).

Where etrue = true strain, strue=true stress, eeng= engg plastic strain, seng= engg stress

Dear Radha

You can measure the true  stress-strain characteristic of an alloy using the UTM by taking simultaneous consecutive photographs of the sample during the  loading.   Then, by comparing the reduction of the diameter of the sample with respect to the original diameter of the sample, this photographic procedure allows you to obtain the instantaneous cross-sectional areas of the sample in order to determine the true stress a long the instantaneous displacement or eventually the true strain.  

Digital Image Correlation (DIC) can be used as answered by Khaled to derive true stress-strain values. One such system is Aramis 3D 5M optical system which can record instantaneous 3D deformation and strain of tensile specimens during test. Please visit the following website for more details.

http://www.gom.com/metrology-systems/aramis.html

Excellent suggestion by Mr.Dinaharan, especially if one is seeking a sub-micrometer scale of the spatial resolution of the displacement of the diameter of the sample. 

To measure the evolution of the cross section during loading you can use a transversal extensometer, the one which is employed to obtain Poisson's ratio. However be careful while using the classical formulae conversion to obtain the true stress-true strain from of the eng. stress strain curve. Indeed they are valid up to necking (UTS) since beyond there is stress triaxiality in the neck (use Bridgeman correction) and, most of the time, damage prior fracture. 

There are several ways to measure the instantaneous cross section including those mentioned above

But the key concern is how to employ the necked geometry data and it's associated non uniform stress, deformation and work hardening to obtain a true Flow stress data

This is not easy nor practical

That is why people prefer other tests such as torsion or compression

The laters have their own limitations as well