Skip to main content
Cross Validated

Return to Answer

remove signature
Source Link
Silverfish
Silverfish
  • 23.9k
  • 28
  • 105
  • 206

I have successfully implemented DTW in 'C' as applied to dynamic signature verification. I used a test data base of Chinese and Dutch signatures to verify EER and got very impressive results. It is currently implemented as a demo on an iPad. My algorithm was hand-coded from several published descriptions. I will share the code if there is a way to get it to you. One thing that also contributed to success was 'normalizing' the input data. This made it a lot easier when comparing disparate data using different sample rates.

Regards, DJ

I have successfully implemented DTW in 'C' as applied to dynamic signature verification. I used a test data base of Chinese and Dutch signatures to verify EER and got very impressive results. It is currently implemented as a demo on an iPad. My algorithm was hand-coded from several published descriptions. I will share the code if there is a way to get it to you. One thing that also contributed to success was 'normalizing' the input data. This made it a lot easier when comparing disparate data using different sample rates.

Regards, DJ

I have successfully implemented DTW in 'C' as applied to dynamic signature verification. I used a test data base of Chinese and Dutch signatures to verify EER and got very impressive results. It is currently implemented as a demo on an iPad. My algorithm was hand-coded from several published descriptions. I will share the code if there is a way to get it to you. One thing that also contributed to success was 'normalizing' the input data. This made it a lot easier when comparing disparate data using different sample rates.

Source Link
user6009837
user6009837
  • 21
  • 1

I have successfully implemented DTW in 'C' as applied to dynamic signature verification. I used a test data base of Chinese and Dutch signatures to verify EER and got very impressive results. It is currently implemented as a demo on an iPad. My algorithm was hand-coded from several published descriptions. I will share the code if there is a way to get it to you. One thing that also contributed to success was 'normalizing' the input data. This made it a lot easier when comparing disparate data using different sample rates.

Regards, DJ