Is this a typo in the paper?

[zzowenzz]

Hi, I have a question about the chapter Approach in the paper with

we estimate a dense displacement field (f1,f2) which maps each pixel (u, v) in I2 to its corresponding coordinates (u′, v′) = (u + f 1(u), v + f 2(v)) in I2

Should it be mapped from I1 to I2 instead of I2 to itself?