Preamble: I don't know anything about photo-editing or signal processing. But I am a composer who happens to be working on a project in which sound / light overlap.

Question: Is it possible to filter a given image through a spectrum of pre-determined wavelengths? (i.e. Passing an image through a filter in which only 480nm, 560nm, 678nm, 714nm are able to pass through? The numbers here are completely arbitrary.)

Many thanks.


If the image is given in the conventional format of sampled pixel values of an sensor matrix, then no it's not possible to apply any software means of EM wavelength/frequency processing on those pixel data. The reason is clear: the sensor pixels are not capturing time variation of electromagnetic wave, they only capture the static (or average) intensity at a point and collectively the intensity distribution along the sensor surface points. Wavelength (and frequency) sensing at a point requires a time domain capture, such as performed by the RF radio antenna or a microphone (sensing the acoustic pressure wave), which therefore enables subsequent frequency/wavelength processing possible.

Therefore practically there remains only a hardware lens filter possbility which is indeed a trusted parctice in the field of photography.

Look for image filters from optics part producers, they can supply any type of filter you want (they are a litle pricey however)


This may be possible to do in software (i.e. after a picture has already been taken), provided the camera captured all the wavelengths reliably in the first place. You can convert the image to hue, saturation, value (HSV) format and then apply a "band pass" filter on the hue dimension which captures the intensity independent color information. Example.

In hardware, this can perhaps be done by placing different filters in front of your camera lens and taking multiple pictures of the scene. I am doubtful if regular photography cameras will work for this application if you are looking for very precise control on spectral content, in which case you'll need to use an image taken with a (hyper)spectral camera. Example.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.