You could use a standard curve drawn by plotting the log of the size of the markers against the migration distance and then derive from it the size of the fragment of interest. However, with closely spaced bands in the marker lane, eyeballing the size of the bands in the sample is usually sufficient. Migration can be affected by the salt concentration of the sample, which limits the accuracy of size determination anyway. Concerning the top gel, even if it "smiles", all bands are aligned across the gel, so they all have the same size and you should use the bands that are closest to th marker lane to make the size estimation.

Use Gel campar, Bionumerics or Phoretix 1D software. They are good software for gel analysis.

Dr. Pierre Béguin i found band size differ by 1-4 bp in most of the cases by deriving log of size of band against the migration distance. Do i need to make general consideration for band scoring or band scoring should strictly based on band length eventhough they differ by 1 -4bp??

Size differences of 1-4 bp cannot be reliably assessed by agarose gel electrophoresis, due to the thickness of the bands. In your case, looking at both gel pictures, it seems that all samples contain fragments having the same length. The fastest migrating band may correspond to primers, if this is an analysis of PCR reactions