| Non-line-of-sight imaging is a new type of computational imaging technology for out-of-sight targets.It can image hidden targets.It has huge application potential in autonomous driving,disaster rescue,military counter-terrorism,and medical imaging.Traditional non-line-of-sight imaging systems consist of an interface,a laser source,and a receiver which measuring the time-of-flight information of the photons scattered by the interface,and the target object is reconstructed by an algorithm.This system requires a pulsed laser and a time-resolving detector.It scans the sampling points on the interface and uses a large amount of data to achieve high-resolution,high-signal-to-noise ratio 3D target reconstruction.This system generally uses a femtosecond pulsed laser as the laser source,and a single-photon avalanche diode or a streak camera as the receiver;it is expensive and the system is complex.In recent years,passive non-line-of-sight imaging technology based on spatial coherence or intensity coherence has reduced system cost,sampling time,and reconstruction time,but has lower resolution and is greatly affected by ambient light.Aiming at the problems of high cost and high complexity of non-line-of-sight imaging systems,this thesis uses a time-of-flight camera instead of a single-photon avalanche diode or streak camera to measure the time-of-flight information carried by scattered photons,effectively reducing the system cost and increasing sampling speed.Aiming at the problem that the non-line-of-sight imaging system is greatly affected by ambient light,which is not conducive to practical application,this thesis uses 4 infrared laser diodes with a wavelength of 850 nm as the light source,which is scattered by frosted glass into a beam,and an infrared filter pole is added in front of the camera lens.The effect of ambient light is greatly reduced,and non-line-of-sight imaging under natural light is realized.Finally,in order to solve the problem of poor resolution caused by system noisein non-line-of-sight imaging systems,this thesis proposes a reconstruction algorithm based on Gaussian fitting,which is obtained by Gaussian fitting of the intensity values corresponding to the same pixel in multiple images optimal value,effectively reducing noise in the image. |
