Inexpensive 3D 'Nano-Camera' for Medical Imaging and Car Collision Detection Systems
A three-dimensional 'nano-camera' that can operate at the speed of light has been developed by a research team in the MIT Media Lab. It could be used in medical imaging and collision-avoidance detectors for cars, and to improve the accuracy of motion tracking and gesture-recognition devices used in interactive gaming. The $500 camera probes the scene with a continuous-wave signal that oscillates at nanosecond periods. This allows the team to use inexpensive hardware – such as off-the-shelf LEDs that strobe at nanosecond periods. The camera is based on time of flight technology like that used in Microsoft's second-generation Kinect device, in which the location of objects is calculated by how long it takes a light signal to reflect off a surface and return to the sensor. "Using the current state of the art, such as the new Kinect, you cannot capture translucent objects in 3-D," says graduate student Achuta Kadambi. "That is because the light that bounces off the transparent object and the background smear into one pixel on the camera. Using our technique you can generate 3-D models of translucent or near-transparent objects."
Transcript
00:00:05 in 1964 doc Edon at MIT used a stroboscopic technique to visualize a bullet as it travel through an apple in 2011 researchers at MIT set their sights on a new goal to capture light as it travel through a scene this technique was called FAL photography and used a half million dollar worth in scientific equipment to achieve this goal we introduced Nan photography which
00:00:31 is a essentially a multi-depth camera that has one application which allows it to visualize light as it travels through a scene time of flight Imaging is an increasingly popular method of range Imaging characterized by high frame rates in a typical implementation a square code modulates the illumination signal the camera intensity measurement is a square wave that is shifted
00:00:54 relative to a reference signal this shift encodes the distance unfort fortunately time of flight cameras fail when there are depth ambiguities for instance in the case of a transparency we consider time of flight in the context of time profile Imaging here a single opaque wall has a spike at a single time
00:01:17 slot for a transparency in front of a wall we observe two spikes one from the transparency and one from the wall behind complex non-sparse time profiles occur in specialized cases such as in that of scattering we consider sparse profiles this can be modeled as a linear combination of impulses at the locations encoding Optical path lengths in our forward model we send a
00:01:45 binary code to the illumination when we measure the return code the shift encodes the path length and distance as before in the presence of a Time profile the illumination code is convolved with the environment profile to provide the measured response by deconvolving the measured cross correlation function we are able to resolve the sparse
00:02:06 profile our implementation is simple requires only a single modulation frequency and can be implemented on commercially available Hardware applications include light sweep Imaging you can observe light hitting the vase and then the stuffed animal behind light then sweeps and hits the lion behind the
00:02:37 Mario and then the back wall observe that the back wall shows through the vase which highlights the multipath capabilities of our camera light leaves first from the specularities on the vase and then from the entire scene we demonstrate additional applications including ranging of transparent objects looking through diffus using material and resolving
00:03:02 multipath edges so what happens is we have a light source and this light source sends a pulse of light to the object in the scene and back to the camera and this camera measures the time it takes for the light to go to the from the light source to the object and back because light travels at a constant speed we can measure the time of flight and therefore
00:03:25 we can calculate the distance that light is traveled therefore we can build up a threedimensional objects are a threedimensional view of the object that we are Imaging now the problem is if we have a transparent object in the scene with the technology that we've developed here we can measure the threedimensional uh we can measure the depth of martini glass
00:03:45 and we can measure the depth of the toy line so the time of flight camera has uh many interesting applications and one of them is transient and ftoc imaging uh so what does it mean it means that we can image things at Ultra fast speed and it it means also that we can make videos of light slices as light traverses so how do we achieve it at such a low cost uh we designed a new camera which has
00:04:12 Hardware capability where light when it interacts with the scene and comes back uh by using ideas from mathematics we are able to uh UNS smear or deconvolve light paths and having done so uh we are able to trace the exact path of light as it travels e