Hi,
  Here is Mavis. I have 2 DCM folders of 1 patient. The folders contain sagittal scan and axial scan. I used Dcm2nii to convert DCM to NII, however, output 2 NII with low quanlity pictures. I wonder if I could input all pictures to 1 NII, so that the pictures of NII will be more accurate. 
  My email address is zmzforever@pku.edu.cn. I appreciate your help.

This is the intended behavior of dcm2niix. It attempts to losslessly convert your DICOM data into NIfTI images. There are many coregistration tools (ANTs, FLIRT, 3dAllineate, SPM, niftyreg, Plastimatch) that are designed to align NIfTI images. This is beyond the scope of dcm2niix.

I suspect if you have clinical scans with good in-plane resolution but thick distance between slices, the results of averaging a sagittal scan and an axial scan will not be acceptable. If you can set up sequences for future acquisitions, you could use the time required to acquire these two different sequences to acquire a single isotropic sequence. In Siemens terminology, 3D sequences like MP-RAGE, SPACE and 3D dark-fluid (FLAIR) sequences provide isotropic resolution with good SNR in a reasonable scan time.