David Smith

Further to Chris Jones' answer, it might be much simpler - if the OP is just asking for a way to differentiate between a white surface and a black surface, the white surface will typically reflect more IR radiation than a black one. So, you can rig up an IR LED shining outwards, and an IR phototransistor or photodiode to measure the reflection. There are various ways to implement the detection circuit, but one way I’ve used is to wire up an IR phototransistor in series with a variable resistor (to provide a sensitivity control), creating a voltage divider (resistor on the top half, phototransistor on the bottom half). I then created a second voltage divider using two equal-value resistors, and then fed these two voltages into a comparator.

As the amount of IR light reflected back to the phototransistor rises, it starts to conduct more, so the voltage produced by the divider drops. When this crosses the threshold (defined by the fixed voltage divider), the comparator switches.

Here's an example circuit (credit: seekic.com):

Infrared (IR) sensors do not detect color - they detect heat energy emitted from object surfaces, typically in long wavelengths of the electromagnetic (EM) spectrum, wavelengths to which the human eye is not sensitive or is blind. With this in mind, dark or black objects like car tires or asphalt absorb more EM energy from sources like the sun than they reflect so they show up as white-hot to IR detector materials; whereas, lighter colors and whites reflect more energy such that objects like chrome trim on cars reflect secondary heat or lack of heat (cold) from sources like the sky (generally cold temperature from space) so these objects appear black to IR sensors. The black-hot/white hot can be remapped/swapped in processing electronics and displayed per user preference

Figure 1 gives a crude block diagram of signal flow converting IR energy to LED display monitor. To simplify the processing and avoid color processing, will assume a grayscale mapping from analog-to-digital converter (ADC) to display where black is cold and white is hot. IR energy at some wavelength, λ is sensed by a large array of detectors, a charged coupled device (CCD) where photonic energy is converted to current, amplified and converted to voltages, multiplexed into a stream and digitized into (say) m-bits, some arbitrarily large instantaneous dynamic range. Ultimately, the m bits and any burst efficiency characteristic in the timing of the front-end electronics multiplexing process is mapped to 8 bits as well as timing for display, because the human eye cannot distinguish more than about 5 bits of grayscale. Also, since typical digital-to-analog converters (DAC) are 8 or 10-bit devices, we map our final dynamic range into what those integrated circuits can provide.

Problems to solve - level and gain non-uniformities.

A non-uniformity problem with detector elements of the CCD array arises which must be solved. Furthermore, in doing so, input sensor stream of m bits can also be mapped to the output display stream of 8 bits. The CCD detectors exhibit both level and gain non-uniformities due to detector material impurities created during the manufacture process. Theoretically, if all detectors are exposed to the same blackbody photonic temperature source, they should all respond with the same current, but in practice, they do not - this constitutes a level non-uniformity for which we must correct. Secondly, as the blackbody source temperature is changed (say, elevated) all detector elements should respond with the same (higher) current, but again, they do not, nor do they respond in the same manner as at the previous (or other) blackbody temperatures - this constitutes gain non-uniformities for which we must also correct.

Calibrate out non-uniformities as a first order effect.

In Figure 2, all the detector element of a CCD are exposed to a uniform blackbody source temperature. (For brevity, only 4 detectors shown using colored dots to depict their ADC values.) Invariably, detectors produce different ADC values, levels, which image processing electronics stores, averages and provides differencing to drive all detector levels to the average, L_Ave.

In Figure 3 after level correction has been applied for the lower blackbody temperature, the blackbody source temperature is raised. Again, the electronics measures the ADC value of all detector elements at the higher temperature, stores them, finds an average, H_Ave, and can now find an average gain, H_Ave - L_Ave and drive gain values for each detector element to match an average gain. A level offset exists from the gain correction so a final, medium blackbody temperature is applied to bring all detectors essentially, to the same L_Ave value.

m-bit to 8-bit mapping.

Now that the CCD detector elements have been equalized by the image processing electronics, an adaptive technique gathers scene statistical information to map m bits to 8 bits. Figure 4 shows a histogram of input dynamic range of 0 to 2^m-1 and number of occurrences for pixel intensities over some dynamic range Min to Max, and a scaled output histogram with dynamic range 0 to 2⁸-1. At this point in the processing chain, some form of a cumulative distribution function, based on the input histogram, can be applied to the forward m-bit video stream via a calculated lookup table to produce an output stream of 8 bits which is converted by DAC to voltages specified by the video standard for display monitor (LED array or whatever).

Summary

Cold IR temperatures in either stream are defined here as low ADC digital values and appear as darker or black pixels in the display, while hot temperatures produce high ADC digital values and are displayed as brighter or white pixels in the display. Again, this can be inverted such that cold IR temperatures are displayed as white and hot IR as black pixel values.