Can I transform colors of scanned pdf files and reduce the scan resolution to save memory keeping an existing text layer from OCR?

I have a pile of pdf files which have been scanned long ago and which are already searchable (i.e. they went through OCR). However the light level and contrast settings were not optimal. **Is it possible to reduce the bits per pixel of the existing files to some reasonable low level** in order to save storage space (make color-curve transformations, posterize or even binarize to black and white like in Gimp or other image-manipulation programs)**?** The files are scanned with 600 dpi and already searchable, i.e. in addition to the scanned image there is a text layer. Probably the scan resolution had been chosen so high in oder to obtain better OCR results. But it makes them excessively large. I think, a scan with 200 dpi would have created good visual quality with much less memory requirements. I want to maintain the OCR generated text layer with its good OCR quality. **What are the proper command?**