How do single-pixel cameras work?

To obtain images using a single-pixel camera we need a Digital Micromirror Device (DMD) because single-pixel cameras only possess a single detector (article).

From my understanding, a single-pixel detector can only detect the intensity of a single spatial point, whilst an array of detectors would be able to recreate the whole image as it would have many detectors because it would detect the intensity of the several points of the picture.

This makes sense, but then I have come across the following image for the experimental setup showing the pattern created by the DMD and the image projected onto the detector.

My limited understanding of the subject makes me think that the image projected onto the THz detector, which is labeled of Collection Optics in the picture, has many pixels instead of a single one.

Shouldn’t the DMD only allow a single-pixel to go through?

I know that my train of thought is probably wrong because it would take ages to create the image because a lot of measurement would be necessary. So where am I going wrong?