Cooling a CCD will reduce its dark current and make it more sensitive to the frequencies it detects. But it will not make it significantly more sensitive to the longer wavelengths that an object near body temperature emits. The maximum wavelength that a normal Si CCD will detect, at any temperature, is about 1.1 microns, which is well below what you need.
You can not make a FLIR by cooling a standard CCD. There is no magic for that. If there was, Fluke and others would be using that for their thermal imagers rather than using expensive bolometer detectors.