The latest motion blur correction method will bring high-quality imaging to dark environment.

　　Since the beginning of photography in the mid-19th century,imaging technology has developed rapidly.With the continuous development of technology,many state-of-the-art cameras for high demand applications believe that their mechanism is very different from that in consumer oriented devices.One of the cameras uses so-called"single photon imaging",which may produce a good dark environment and fast dynamic scenes.Let's look at the difference between single photon imaging and traditional imaging.

　　On June 9,2021,Tokyo-researchers at Tokyo Institute of technology introduced a method to solve the limitations of single photon imaging:motion.A team led by Professor Hiroyuki Hamamoto has developed an algorithm that can solve the blur caused by the motion of the imaging object or the common blur of the whole image,such as the blur caused by camera jitter.

　　When a CMOS camera(such as a camera in a smartphone)is used to capture images,moving objects are usually imaged by shortening the exposure time.Image crazy single photon camera(single photon imaging)consists of a series of very short individual exposures that can capture continuous frames.Each of these frames consists of a binary grid of 0 and whether the photons are received during the exposure.

　　However,the speed of single photon imaging is much higher than that of CMOS camera.Due to its completely digital characteristics,single photon imaging allows ingenious reconstruction algorithms,which will be constructed for technical constraints or difficult scenes.

　　The team's method solves many limitations of the existing deblurring technology of single photon imaging.When there are multiple moving objects in the image scene and move and/or overlap each other at different speeds,the technology will generate low-quality images.This strategy uses a more dynamic strategy,rather than adjusting the whole image according to the estimated motion of an object,or adjusting the whole image according to the idea of considering the spatial region of object motion.The motion estimation algorithm tracks the motion of a single pixel by statistically evaluating the change of bit value over time(on different bit planes).Experiments show that in this method,the motion of a single object is usually accurately estimated.

　　After obtaining information from the initial algorithm,the second deblurring algorithm groups pixels with the same motion together to identify individual objects moving at different speeds in each bit plane.Therefore,each region of the image can be blurred independently consistent with the motion of the object experiencing it.The simulation shows a clear and high-quality image,even in a weak light dynamic scene containing multiple objects moving at different speeds.

　　"The method of obtaining clear images with limited photons is very useful in many fields such as medicine,safety and science,"Hamamoto said."Our approach is expected to bring new technologies to achieve high-quality imaging and ultra slow recording in dark environments such as space,which will far exceed the capabilities of today's fastest cameras."