Most of Ippmethods for 8 bit is 'unsigned' and for 16bit is 'signed'. Is there any specific reason for choosing this? We mainly work on 8u and 16u, but most of the methods for 16bit is signed. So we would like to knowwhy it is not supported or is there any workaround, other than upgrading to 32s or 32f?
For more complete information about compiler optimizations, see our Optimization Notice.