Denoising of the distance data from underwater time-of-flight (tof) cameras

In underwater sensing, there is a demand for high-speed, high-resolution optical range imaging of objects in turbid water. Our prior work adapted commercially available timeof-flight (ToF) hardware to function in the unique optical conditions imposed by the water mediumnamely, high levels of signal absorption and scattering. In this paper, we propose an algorithm to denoise the distance data captured with a ToF camera in turbid water conditions. To this effect, we use a 2-parameter scale-space generated by the iterative nonlocal means filter to denoise the DCS data. We then use these denoised DCS data to reconstruct the denoised distance image. We test the algorithm on data obtained from a simulated optical channel as well as real experimental ToF data. Compared with the local denoising methods, the results obtained with the proposed algorithm provide demonstrably improved distance images.