Introduction

The ultrafast photography, which can acquire sequential images of ultrafast dynamic phenomena with a high temporal resolution, plays an important role in the fields of biomedicine, chemistry, and physics1,2,3,4,5,6,7,8,9,10,11. In recent years, the single-shot ultrafast imaging technologies have been developed, including the sequentially timed all-optical mapping photography (STAMP)12, the compressed ultrafast photography (CUP)13 and others14,15,16,17,18,19,20. In these techniques, the STAMP and CUP can raise the temporal resolution to hundreds of femtoseconds. However, the STAMP was not suitable for observing self-emission objects and the CUP needs the image reconstruction with compressed-sensing algorithms. The image sampling strategy based on the streak camera can also acquire sequentially time-resolved two-dimensional (2D) images21,22,23.This technique used the pinhole array or the fiber bundle to convert a 2D image to a set of one-dimensional (1D) images, which were then detected by the streak camera. The sequential 2D images with about 10 ps temporal resolution can be reconstructed under the spatial relationship between the 2D and 1D images. Due to only one column of image signal being sampled by the streak camera, the field of view of the imaging system is limited due to the limited quantity of effective pixels. Therefore, in order to enlarge the field of view for recording optical images, a parallel multi-group sampling strategy has been proposed. In this strategy, several columns of 1D images are parallelly sampled by a streak camera to enlarge the quantity of effective recorded pixels. Though the quantity of pixels has been increased, the recording time length has been decreased without decreasing the temporal resolution. Therefore, the proposed strategy has provided a method for balancing the quantity of pixels and the recording time length. In this paper, the demonstration experiment of parallel multi-group sampling has been performed. In the experiment, a fiber bundle was used to convert the 2D image into four groups of 1D images, each of which was placed at a specific area and detected by a streak camera. A proof experiment was also carried out for observing the flight of light.

Experimental setup

The ultrafast photography system based on the parallel multi-group sampling strategy is shown in Fig. 1. The fiber bundle was a sampling structure, which was composed of one square area with 64\(\times\)64 pixels in the input end and four columns with 1\(\times\)1024 pixels of each one in the output end. The fiber bundle can be arranged as any m\(\times\)n pattern according to specified needs, m and n are integers. However, under the condition of a certain value of m\(\times\)n, we will have a smaller framing number when m is larger and we will have a lower spatial resolution when n is larger that is limited by the performance of the streak camera. In our design, the pixel number of 1024 was slightly larger than the spatial resolution ability of nominal streak cameras, about 500 pixels. In this way, the spatial resolution ability of the streak camera can be fully used. Each pixel of the fiber sampling structure consisted of 4\(\times\)4 fibers, the diameter of which was about 20 μm. For the current fiber bundle each unit of 256\(\times\)4 fibers was stacked at the input end and arranged with head to tail at the output end, respectively. Therefore, the dimension of the square area in the input end and each column in the output end was about 5.12\(\times\)5.12 \(\hbox {mm}^2\) and 0.08\(\times\)81.92 \(\hbox {mm}^2\), respectively. The fiber structure was fabricated with following steps. A fiber bundle array with 256\(\times\)4 fibers was firstly fabricated with an equipment. Then, each fiber bundle array was artificially arranged according to a specific sequence in the input and output ends. It means that the input end of the fiber bundle and the output end of each column is composed of 256\(\times\)256 fibers and 4\(\times\)4096 fibers, respectively. In the proof experiment, the optical object was a plane with a mark of NINT, which was illuminated by a laser with 532 nm wavelength and 30 ps pulse duration time (full width half maximum, FWHM). The plane was located at about 165° relative to the incident laser direction, leading to different arrival times of light along the plane surface. The dynamic phenomenon of the flight of light was imaged onto the input end of the fiber bundle through a lens with 85 mm focal length. The output end of the fiber bundle was coupled to the streak camera through a lens with 100 mm focal length. The streak camera was developed by Xi’an Institute of Optics and Precision Mechanics of Chinese Academy of Sciences. Its features are an 8\(\times\)8 \(\hbox {mm}^2\) size photocathode and about 4 ps temporal resolution. The image obtained by the streak camera was composed of four areas, each of which represented an 1D image evolving with time. It should be noted that the duration time of the pulsed phenomenon cannot be exceed the scanning time represented by each 1D image streaking area to avoid the overlapping of the four 1D streaking images.

Fig. 1
Fig. 1
Full size image

Experimental setup of the ultrafast photography system for observing the flight of light.

Results and discussion

The spatial relationship of each pixel between the input and output end was acquired by analyzing the image which was obtained through placing a knife edge on the input end of the fiber bundle. The knife edge was first adjusted to be nearly perpendicular to each unit of 256\(\times\)4 fibers. Then the knife edge was moved slowly with a translation stage to allow only the head part of each unit of 256\(\times\)4 fibers, representing 64\(\times\)1 pixels, to be illuminated. During the movement process of the knife edge, the streak camera continuously captures images, which were displayed in real-time on the computer screen. At the beginning, the input end was completely blocked. Then, the knife edge was moved until a bright spot of each unit of fiber bundles was observed by the streak camera. Therefore, the position of the head of each unit can be determined with the positions of the bright spots in the obtained image of the streak camera. In order to demonstrate the spatial relationship, the plate with the mark of NINT illuminated with a stable light source was imaged with the photography system. The raw image was recorded by the streak camera at the static state, which is shown in Fig. 2a. The 2D image was recovered based on the spatial relationship between the input and output end of the fiber bundle and is shown in Fig. 2b. The position relation of pixels between the 1D image and 2D image is shown in Fig. 2c. It should be noted that the median filtering and the flat field correction was performed on the recovered image to improve the image quality and remove the spatial uniformity, respectively. The result, which can roughly identify the mark of NINT, indicates the 2D imaging capacity with this sampling principle based on the current photography system.

Fig. 2
Fig. 2
Full size image

(a) The raw image obtained with the photography system under the condition that the streak camera was operated in the static state and the image object was a plate with a mark of NINT illuminated with a stable light source. The scanning direction of the streak camera is horizontal. (b) The recovered image with the mark of NINT from the raw image. (c) The position relation of pixels between the 1D image and 2D image.

The spatial resolution of this technique is determined by the fiber, the lens and the streak camera. The spatial resolution of the fiber is roughly equal to its diameter. For the current system, it is about 20 μm. The ultimate spatial resolution of an ideal lens can be expressed as 0.61\(\lambda\)/n/sin\(\theta\) , where \(\lambda\) is the light wavelength, nsin\(\theta\) is the numerical aperture. For the current system, it is about 9 μm. The spatial resolution of the streak camera is about 150 μm at the output end of the fiber bundle, assuming that the streak camera can resolve about 500 pixels in the Y-axis direction. Therefore, the spatial resolution of the system is mainly limited by the streak camera. For the current system, an experiment has also been carried out to evaluate the spatial resolution of the imaging system by analyzing the edge spread function (ESF) of the image of the fiber bundle output end captured by the streak camera at the static state, which is shown in Fig. 3a. Fig. 3a is the labelled area of Fig. 2a. The output end of the fiber bundle was also observed with a microscope, which is shown in Fig. 3b, to verify the arrangement of optical fibers. The four columns of fibers were stacked tightly to each other to form the unit of 256\(\times\)4 fibers. Each side of the unit can be seen as an edge, which can be used to evaluate the spatial resolution of the streak camera. The ESF can be roughly obtained by averaging 60 samples along the vertical direction in the stamped region of Fig. 3a, which is shown in Fig. 3c. The light intensity distribution at the output end of the fiber bundle is also theoretically calculated in Fig. 3c to evaluate the spatial resolution deterioration. The spatial resolution of the streak camera at the output end of the fiber bundle is indicated to be about 0.60 mm by analyzing the distance between the 10% and 90% of the peak intensity, which is a mean value calculated with the left and right ESFs. Consequently, the performance of the imaging system is mainly limited by the spatial resolution characteristic of the streak camera.

Fig. 3
Fig. 3
Full size image

(a) The image of the fiber bundle output end captured by the streak camera at the static state. (b) The output end of the fiber bundle observed by a microscope. (c) The spatial resolution deterioration analysis by evaluating the ESFs.

The ability of recording sequentially time-resolved 2D images with this photography system was also demonstrated by observing the phenomenon of the flight of light. The experimental setup is shown in Fig. 1. The laser with 532 nm wavelength and 30 ps duration time (FWHM) illuminated the plate with the mark of NINT. Due to the incident angle between the laser and the plate surface, the plate was sequentially illuminated by the laser from the side of T to the side of N, which is attributed to the flight of light. The raw image recorded by the streak camera at the dynamic state was shown in Fig. 4. The scanning direction of the streak camera is horizontal and the time on the left is earlier than the time on the right.

Fig. 4
Fig. 4
Full size image

The raw image obtained with the photography system under the condition that the streak camera was operated in the dynamic state and the image object was a plate with a mark of NINT illuminated with a laser source. The scanning direction of the streak camera is horizontal.

The sequential frames were recovered from the raw image by using the spatial relationship of each pixel of the fiber bundle. The median filtering and the flat field correction was also performed on the recovered frames. The sequential recovered frames are shown in Fig. 5. Due to the limited spatial resolution of the streak camera operating at the dynamic state, the marker of “NINT” cann’t be fully identified. However, the dynamic process of the flight of light can be clearly observed from these framing images (see Visualization1). The screen plate was firstly illuminated at the side of the marker “T” and then sequentially illuminated at other letters. In view of the incident angle of 165° between the laser and the plate, the light speed in the air was evaluated to be about 2.9\(\times 10^{8}\) m/s by the comparison of the two fames with the time marker of 21.4 ps and 128.4 ps. The above results indicate that the ultrafast photography based on the parallel multi-group sampling strategy has the ability to observe the ultrafast dynamic process and can be a promising tool in investigating the ultrafast phenomenon.

Fig. 5
Fig. 5
Full size image

The sequential frames obtained with the ultrafast photography system, which indicated the dynamic process of the flight of light (see Visualization1).

This technique can be used to investigate the ultrafast dynamics of self-emission objects in the plasma science or the fluorescent material area. Moreover, it can also be used to observe the ultrafast active objects illuminated under a light with no temporal resolution in biomedicine. In addition, this technique can be improved by optimizing the arrangement of fibers. For example, the fiber bundle can be rearranged into more 1D lines to improve the spatial resolution feature. Furthermore, the direct way to improve this technique is to raise the characteristics of the streak camera.

Conclusion

In summary, an ultrafast photography system, which can record sequentially time-resolved images, has been developed. This imaging system employs a fiber bundle to sample the 2D image and rearranges the image pixels to four lines. The four 1D images were then recorded by a streak camera. The 2D sequential frames can be recovered from the recorded image based on the spatial relationship between the input and output port of the fiber bundle. The demonstration experiment has been successfully performed for observing the dynamic phenomena of the flight of light, indicating this ultrafast photography scheme has the ability of investigating ultrafast phenomena.