I have an imaging system that produces a signal in a pixel, in part, by responding to incident radiance $L(\lambda)$ that passes through a filter with transmission $T(\lambda)$. Lumping all the other parameters under a generic constant I can write the signal: $S = k\cdot\int L(\lambda)\ast T(\lambda)d\lambda\,$. I'm trying to find a way to determine, or even estimate $L(\lambda)$. The system is such that I know $S, k$, and $T(\lambda)$. Is there a way to recover the incident radiance? Note, I am not necessarily looking for a closed form solution, approximation is fine.

  • $\begingroup$ Are S and L functions of time? Your $\ast$ isn't making sense and you didn't name the source equation. $\endgroup$ Aug 12, 2020 at 13:33
  • $\begingroup$ S and L are not functions of time. L, the radiance, is a function of wavelength. It's convolved with the transmission T of a color filter and summed over all wavelengths to produce the signal S (pixel value), expressed in DN in the final digital image. Not sure what you mean by source equation. $\endgroup$ Aug 12, 2020 at 14:20
  • $\begingroup$ This might help: dsp.stackexchange.com/questions/65977/… $\endgroup$ Aug 12, 2020 at 14:50
  • $\begingroup$ So $L(\lambda) * T(\lambda) = \int_{-\infty}^\infty L(\lambda - l) T(\lambda) dl$? $\endgroup$
    – TimWescott
    Aug 12, 2020 at 18:54
  • $\begingroup$ No, perhaps I'm using convolution incorrectly here. The functions are multiplied together. For instance, at 500nm let's say the radiance is 5 $(W/m^2Sr)$, and the filter transmission at that wavelength is 0.5, so the signal recorded is $2.5\cdot k$ $\endgroup$ Aug 12, 2020 at 20:27


Your Answer

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

Browse other questions tagged or ask your own question.