Stray Light Suppression Design and Test for the Jilin-1 GF04A Satellite Remote Sensing Camera

Abstract

The stray light suppression design aims to reduce the impact of stray light on optical systems. For high-resolution optical remote sensing systems, practical tests of stray light suppression performance are essential to ensure optimal functionality. However, due to system complexity and spatial constraints, physical test methods for evaluating the stray light suppression performance of large-aperture, long-focal-length remote sensing cameras remain scarce. To address this issue, a comprehensive test is conducted on the stray light suppression performance of the Jilin-1 GF04A satellite remote sensing camera by integrating multiple test methods, including the environmental light effect test, neighborhood point source response test, key surface response test, and sneak path of stray light test. The experimental results indicate that the stray light response ratios obtained from different test methods are all below 1%. The on-orbit performance of GF04A further validates the effectiveness of its stray light suppression design.

1. Introduction

High-resolution Earth remote sensing cameras are high-performance optical systems specifically designed for Earth observation. They are widely used in fields such as land resource monitoring, environmental protection, urban planning, disaster assessment, and agricultural resource management [1]. To reduce manufacturing and launch costs, miniaturization is one of the development trends for remote sensing cameras. However, to enhance spatial resolution and improve light collection efficiency, cameras need to have large apertures and long focal lengths [2]. The Jilin-1 GF04A satellite, developed by Chang Guang Satellite Technology Co., Ltd., Changchun, China, is the world’s lightest operational optical satellite with a resolution of 0.5 m [3]. Equipped with a coaxial three-mirror anastigmat (TMA) remote sensing camera, it achieves sub-meter spatial resolution, featuring high-efficiency data acquisition, lightweight design, and multispectral observation capabilities.

For high-resolution and large-aperture remote sensing cameras, stray light is a critical factor affecting image quality. Stray light refers to light entering the optical system but propagating outside the designed optical path [4]. After reflection, refraction, or scattering, this light reaches the detector, forming background noise. This results in a degradation of the modulation transfer function (MTF), a reduction in the signal-to-noise ratio (SNR), and a significant impact on radiometric accuracy. In regions with low radiance targets, stray light may even overwhelm the target signals [5,6,7]. Therefore, stray light suppression is a crucial aspect of high-resolution remote sensing camera design, and systematically evaluating stray light suppression capability is key to ensuring the effectiveness of the design.

The methods for evaluating the stray light suppression capability of an optical system can be categorized into analytical and experimental approaches. The analytical method ensures that the stray light level remains within an acceptable range during the design phase through software simulations. The experimental method, on the other hand, demonstrates the actual stray light suppression capability of the system through test. Experimental methods include the point source transmittance (PST) method and the surface source method. The PST method [8,9,10] and the Spatial Point Source Transmittance (SPST) method [11,12] evaluate the stray light level of an optical system both outside and within the field of view, and are widely used in high-precision stray light tests.

On the other hand, the surface source method is used to measure the veiling glare index (VGI). Specifically, it involves testing the illuminance of black and white targets on the image plane of the optical system. This method provides a comprehensive evaluation of the optical system’s stray light suppression capability. For small-aperture optical systems, the surface source method is favored due to its simple test equipment and ease of implementation [13,14]. However, for long-focal-length, large-aperture optical systems, the surface source method requires an integration sphere of infinite size and a black spot positioned at an infinite distance to approximate a 2π steradian extended light source, significantly increasing the spatial size and cost of the test equipment. Clermont et al. proposed the concept of the Stray Light Entrance Pupil (SLEP) to define the limiting entrance pupil for lights that generate stray light when reaching the detector [15]. When the SLEP is significantly smaller than the instrument aperture, scanning its envelope can reduce test time and help minimize the required light source size.

Stray light affects both the MTF and radiometric accuracy of remote sensing cameras. In this paper, the analysis and design of stray light suppression are presented to ensure high MTF and the radiometric accuracy of the GF04A satellite remote sensing camera. To avoid the use of oversized test equipment and reduce system complexity, we integrate four stray light test methods, including environmental light effect (ELE) test, neighborhood point source response (NPR) test, key surface response (KSR) test, and sneak path of stray light (SPS) test. These test methods are used to systematically and comprehensively evaluate the stray light suppression capability of the GF04A satellite, ensuring the achievement of high MTF and high radiometric accuracy standards.

2. Optical System Design

A Korch-type Three Mirror Cassegrain system is designed for the satellite’s high-resolution payload, using Ansys Zemax OpticStudio 2023 R1 pro, as shown in Figure 1. The optical system features a 500 mm aperture and a cross-track field of 1.6°. The aperture stop is set at the primary mirror, and the intermediate image between the secondary mirror and the tertiary mirror is located after the central hole of the primary mirror. The final image is relayed by the tertiary mirror.

At the intermediate image plane, the field stop can be set to exclude the stray light from outer source. Meanwhile, in the optical path, the exit pupil is accessible, in which the Lyot stop can be located for eliminating the scattered stray light from inner surfaces.

The optical system’s MTF is shown in Figure 2, with a 450~700 nm wavelength for panchromatic band, and with the obscuration effect of the secondary mirror baffles considered. The average MTF value for all fields is about 0.25 at the Nyquist frequency 57.14 lp/mm (the pixel size is 8.75 μm).

The optical system’s RMS wave-front error for cross-track fields is shown in Figure 3a, at the interferometer test wavelength (He-Ne laser, λ = 632.8 nm). The values are much lower than the diffraction limit and the maximum is 0.03 λ, between 0.5° and 0.6°.

The relative distortion of the proposed optical system is shown in Figure 3b. It shows that the maximum distortion is only 0.023% at the marginal field.

3. Stray Light Analysis and Suppression Design

In the optical design phase, stray light analysis is a critical step to ensure the effectiveness of the optical system. The energy propagation of stray light can be described as the transfer of non-imaging rays between the emitter and the receiver [16]. The stray light radiative flux $\mathrm{d}{\mathsf{\Phi }}_{\mathrm{c}}$ on the receiver surface $\mathrm{d}{\mathrm{A}}_{\mathrm{c}}$ is calculated as follows:

$$\mathrm{d}{\mathsf{\Phi }}_{\mathrm{c}}=\mathrm{d}{\mathrm{A}}_{\mathrm{s}}·{\mathrm{L}}_{\mathrm{s}}·\cos {\mathsf{\theta }}_{\mathrm{s}}·\mathrm{d}{\mathsf{\Omega }}_{\mathrm{c}}=\mathrm{d}{\mathrm{A}}_{\mathrm{s}}·{\mathrm{L}}_{\mathrm{s}}·\cos {\mathsf{\theta }}_{\mathrm{s}}·{\displaystyle \frac{\mathrm{d}{\mathrm{A}}_{\mathrm{c}}·\cos {\mathsf{\theta }}_{\mathrm{c}}}{{\mathrm{R}}_{\mathrm{sc}}^2}}$$

(1)

where $\mathrm{d}{\mathrm{A}}_{\mathrm{s}}$ and $\mathrm{d}{\mathrm{A}}_{\mathrm{c}}$ represent the differential surface areas of the emitter and receiver, respectively; ${\mathrm{L}}_{\mathrm{s}}$ is the radiance of the emitter; ${\mathsf{\theta }}_{\mathrm{s}}$ and ${\mathsf{\theta }}_{\mathrm{c}}$ are the angles between the normals of the emitter and receiver surfaces and the line connecting their centers; ${\mathrm{R}}_{\mathrm{sc}}$ is the distance between the centers of the two surfaces; and ${\mathsf{\Omega }}_{\mathrm{c}}$ is the solid angle of the receiver surface relative to the emitter surface.

The above equation can be simplified as follows:

$$\mathrm{d}{\mathsf{\Phi }}_{\mathrm{c}}=(\cos {\mathsf{\theta }}_{\mathrm{s}}·{\displaystyle \frac{\mathrm{d}{\mathrm{A}}_{\mathrm{c}}·\cos {\mathsf{\theta }}_{\mathrm{c}}}{{\mathrm{R}}_{\mathrm{sc}}^2}}) ·{\displaystyle \frac{{\mathrm{L}}_{\mathrm{s}}}{{\mathrm{E}}_{\mathrm{s}}}}·{\mathrm{E}}_{\mathrm{s}}·\mathrm{d}{\mathrm{A}}_{\mathrm{s}}=\mathrm{GCF}·\mathrm{BSDF}·\mathrm{d}{\mathsf{\Phi }}_{\mathrm{s}}$$

(2)

where $\mathrm{GCF}=\cos {\mathsf{\theta }}_{\mathrm{s}}·\mathrm{d}{\mathrm{A}}_{\mathrm{c}}·\cos {\mathsf{\theta }}_{\mathrm{c}}/{\mathrm{R}}_{\mathrm{sc}}^2$ represents the projected solid angle of the receiver surface with respect to the emitter surface and is referred to as the geometric composition factor. The bidirectional scattering distribution function ($\mathrm{BSDF})={\mathrm{L}}_{\mathrm{s}}/{\mathrm{E}}_{\mathrm{s}}$ is the radiance of a scattering surface, normalized by the irradiance incident of the surface. $\mathrm{d}{\mathsf{\Phi }}_{\mathrm{s}}$=${\mathrm{E}}_{\mathrm{s}}·\mathrm{d}{\mathrm{A}}_{\mathrm{s}}$ is the radiative flux of the emitting surface, and ${\mathrm{E}}_{\mathrm{s}}$ is the irradiance of the emitting surface.

In Equation (2), to reduce the stray light radiative flux $\mathrm{d}{\mathsf{\Phi }}_{\mathrm{c}}$ at the receiving surface, three approaches can be taken:

• Reduce $\mathrm{GCF}$ by introducing baffles and other blocking elements to obstruct stray light transmission paths.
• Decrease $\mathrm{BSDF}$ by treating the optomechanical structure surfaces to enhance absorption and reduce reflectivity.
• Reduce $\mathrm{d}{\mathsf{\Phi }}_{\mathrm{s}}$ by minimizing the emitted energy from stray light sources.

However, the BRDF of the component surfaces and the stray light flux from the radiation source are difficult to reduce to zero, while the GCF can be potentially reduced to zero [17]. Therefore, adding baffles or other light-blocking elements to the non-imaging light propagation path to reduce the GCF is the most direct and effective method for suppressing stray light.

Considering the analysis above, a stray light suppression design has been implemented in the optical system, as shown in Figure 4. To block the propagation of stray light and reduce the $\mathrm{GCF}$, some baffles are installed. The external baffle is positioned at the outermost part of the optical system and primarily blocks strong light sources outside the camera’s field of view, such as sunlight or Earth-reflected light, from directly entering the camera. Inside the optical system, the internal baffle absorbs and blocks stray light from outside the field of view, as well as most scattered stray light within the cavity between the primary and secondary mirrors, preventing it from entering the subsequent optical path. The secondary mirror baffle is designed to block stray light that bypasses the primary and secondary mirrors but is reflected by subsequent mirrors onto the focal plane. To further enhance stray light suppression, the Lyot stop restricts the entrance pupil beam of the camera, allowing only light reflected by the primary mirror to pass through while blocking light reflected by structural components surrounding the primary mirror. Additionally, baffles are installed near the Lyot stop to directly block light from outside the imaging field of view and non-imaging light that could illuminate structural components and propagate onto the focal plane.

The stray light suppression capability of the Jilin-1 GF04A camera is then verified through software simulations. The method involves the following steps: a finite-size surface light source is located at different positions on the camera’s focal plane. A receiving surface, covering the entire system entrance, is set at the system’s entrance to collect the total energy that reaches the image plane. This energy includes both the effective imaging light and the stray light incident on the image plane. Additionally, a second receiving surface is placed far from the system, sized to match the field of view corresponding to a small focal plane element, ensuring that the collected light consists only of effective imaging light. The VGI is calculated using the following equation:

$$\mathrm{V}={\displaystyle \frac{\mathrm{Stray} \mathrm{light} \mathrm{energy}}{\mathrm{Total} \mathrm{light} \mathrm{energy}}}={\displaystyle \frac{\mathrm{Total} \mathrm{light} \mathrm{energy}\text{-}\mathrm{Effective} \mathrm{light} \mathrm{energy}}{\mathrm{Total} \mathrm{light} \mathrm{energy}}}\times 100\%$$

(3)

The number of rays used for tracing is set to 3,000,000, with a tracing threshold of 1e-6. The results are shown in

Table 1. Results of VGI calculation.

No.	Focal Plane X Coordinate (mm)	Focal Plane Y Coordinate (mm)	VGI
1	0	−70.65	0.70%
2	13.44	−70.65	0.69%
3	40.32	−70.65	0.75%
4	77	−70.65	0.83%
5	117.32	−70.65	0.62%

.

The simulation results show that the system’s VGI is better than 1%, indicating that the system has a good stray light suppression performance.

The PST is an important indicator for measuring the stray light suppression performance of an optical–mechanical system. It is defined as the ratio of the irradiance at the image plane, caused by light rays emitted from a stray light source with a field of view angle θ after passing through the optical system, to the irradiance of the source perpendicular to that direction, i.e.:

$$\mathrm{PST}={\displaystyle \frac{{\mathrm{E}}_{\mathrm{d}}\left(\mathsf{\theta }\right)}{{\mathrm{E}}_{\mathrm{i}}\left(\mathsf{\theta }\right)}}$$

(4)

In the software simulation, the light source is positioned at the entrance of the external baffle of the Jilin-1 GF04A camera, filling the entire camera aperture. The radiance of the light source is set to 1 W/m², with a total of 500,000 rays, and the light power threshold is set to1 × 10⁶. During the simulation, the light source is continuously rotated while the detector’s response to rays is recorded at different incident angles, with the results shown in Figure 5. The reflectivity and scattering rates of the mirror, baffles, and rear cover are increased, exceeding the actual reflectivity and scattering rates.

As shown in Figure 5, with the incident angle increasing, the PST in all directions shows a stable decreasing trend. When the incident angle exceeds 5°, the PST drops below 1 × 10^-4. The simulation results indicate that the stray light suppression is highly effective.

4. Test Methods and Results

To test and verify the actual stray light suppression capability of the optical system, several test methods are proposed, as shown in

Table 2. Stray light test methods.

Test Method	Test Purpose	Description
Environmental Light Effect (ELE) test	Check for strong stray light far beyond the field of view	Test stray light in a wide-angle range along and perpendicular to the track
Neighborhood Point Source Response (NPR) test	Check for strong stray light near the field of view	Test stray light in a small angle range near the field of view
Key Surface Response (KSR) test	Check for strong stray light that does not pass through key surfaces	Cover the secondary mirror and test the intensity of stray light reaching the image plane
Sneak Path of Stray light (SPS) test	Check the gaps in the camera back cover	Close the light inlet and test whether the external light source can be sensed by the focal plane detector

. These methods are summarized and improved based on existing approaches and our project experience, enabling a targeted test of critical stray light in remote sensing cameras.

4.1. ELE Test Result

The ELE test is used to simulate the disturbance caused by strongly reflective objects outside the field of view (such as clouds or snow-covered mountains) on the imaging of low-reflectivity target areas (such as forest vegetation) by high-resolution remote sensing cameras. The test system consists of a 1.2 m-diameter integration sphere (Hefei, China, Anhui Institute of Optics and Fine Mechanics, same below), a crane, the camera under test, and ground inspection equipment. The test setup is shown in Figure 6.

The ELE test steps are as follows:

• To begin with, the integration sphere is turned on and its brightness is adjusted to a high value.
• After positioning the integration sphere directly in front of the camera, the camera entrance is kept open and the camera’s response $\mathrm{D}{\mathrm{N}}_{1\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$ is recorded.
• Next, the camera is placed at the side of the integration sphere. Under these conditions, the camera position is adjusted so that the angle between the normal line of the integration sphere and the camera’s optical axis changes, and the camera’s response $\mathrm{D}{\mathrm{N}}_{2\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$ at different orientations is recorded.
• Afterwards, when the camera is positioned at different orientations, the light source of the integration sphere is turned off, and the camera’s response $\mathrm{D}{\mathrm{N}}_{3\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$ under dark-field conditions is recorded.

During the whole test, the camera operates in a low-light mode (60× speed reduction). Detectors (CCD1~5) are used to detect the incident light signals and record the response data. The stray light response ratio ${\mathrm{V}}_1$ at different deviation angles is calculated using the following formula:

$${\mathrm{V}}_1=(\mathrm{D}{\mathrm{N}}_{2\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}-\mathrm{D}{\mathrm{N}}_{3\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}})/\mathrm{D}{\mathrm{N}}_{1\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$$

(5)

Through data calculation, the stray light response ratio at different deviation angles in the along-track direction is presented in

Table 3. Stray light response ratio ${\mathrm{V}}_1$ at different deviation angles in the along-track direction.

Angle (°)	B1	B2	B3	B4	P
30	0.0984%	0.0863%	0.1034%	0.1132%	0.0789%
40	0.0651%	0.0771%	0.0863%	0.1064%	0.0587%
50	0.0436%	0.0369%	0.0269%	0.0364%	0.0321%
60	0.0286%	0.0277%	0.0357%	0.0479%	0.0335%
70	0.0311%	0.0274%	0.0262%	0.0257%	0.0249%

.

In Table 3, B1, B2, B3, B4, and P represent different spectral bands. Furthermore, stray light response ratio at different deviation angles in the cross-track direction is shown in

Table 4. Stray light response ratio ${\mathrm{V}}_1$ at different deviation angles in the cross-track direction.

Angle (°)	B1	B2	B3	B4	P
30	0.0674%	0.0672%	0.0786%	0.0532%	0.0578%
40	0.0359%	0.0333%	0.0344%	0.0464%	0.0469%
50	0.0335%	0.0291%	0.0278%	0.0367%	0.0386%
60	0.0286%	0.0345%	0.0268%	0.0371%	0.0371%
70	0.0326%	0.0249%	0.0325%	0.0258%	0.0287%

.

The results show that the stray light response ratios ${\mathrm{V}}_1$ tested at large angles in both the along-track and cross-track directions are both less than 0.12%, and they monotonically decrease with increasing deviation angle, indicating that stray light outside the field of view has been effectively suppressed.

4.2. NPR Test Result

The optical system can converge light within the imaging field of view onto the image plane, while light near the imaging field may either converge outside the detector’s photosensitive area or be intercepted by structures such as baffles. This intercepted light can easily form ghost images or other types of light spots on the detector’s photosensitive area after a single reflection. The NPR test involves a camera placed on a turntable which receives the light emitted by a collimator. The response in small angular ranges is analyzed to detect subtle ghost images or strong light scattering phenomena.

The NPR test system consists of a star target with a diameter of approximately 0.2 mm, a light source composed of a halogen lamp and an integration sphere, a 30 m collimator (Nanjing Astronomical Instruments Co., Ltd., Nanjing, China), a one-dimensional turntable (Beijing Institute of Aerospace Control Devices, Beijing, China), black cloth, the camera under test, and ground inspection equipment, as shown in Figure 7.

Firstly, the camera is positioned in a direction perpendicular to the detector array (long edge) and the turntable, with the camera’s pitch angle adjusted to ensure that the collimator is accurately aligned with the camera under test. Then, the turntable is rotated clockwise at an angular speed of 0.05°/s, during which the direction of image movement must remain consistent with the integration direction. Meanwhile, the brightness of the integration sphere is regulated to keep the DN value of the star target during the scanning process at a high level (close to saturation or fully saturated). Next, the corresponding turntable angle ω is estimated when the star point is captured in each spectral band. Finally, the turntable is rotated at a constant speed, continuously collecting 100,000 lines of images to guarantee the integrity and usability of the experimental data. The star point image in the P-spectrum is shown in Figure 8.

No abnormal bright spots are found above or below the star points in the image of each spectral band. The DN values at the pixels in the same column as the center of the bright spot are shown in Figure 9, where the x-axis represents the row number and the y-axis represents the DN value. There are no bright spots resembling ghost images above or below the star point image.

4.3. KSR Test Result

There are various paths for a stray light incident on the focal plane of an optical system. For reflective systems, stray light bypassing certain key imaging elements can have a particularly severe impact. The KSR test involves blocking a specific critical surface of the optical system, measuring the illuminance or response value on the image plane, and comparing it to the response value under dark-field conditions.

The key surfaces in this optical system are the secondary mirror and the Lyot stop. The camera’s entrance is aligned with the center of a 1.2 m-diameter integration sphere. With the integration sphere turned on, the veiling glare index ${\mathrm{V}}_2$ of the system with the secondary mirror covered is tested by recording the imaging DN values before and after covering the secondary mirror. Similarly, the veiling glare index ${\mathrm{V}}_3$ of the system with the Lyot stop covered, which indicates the proportion of stray light generated by the structure in front of the Lyot stop, is tested by recording the imaging DN values before and after covering the Lyot stop. The test setup is shown in Figure 10.

The steps for KSR test are as follows:

• To start with, all baffles are installed on the camera. The camera parameters and the energy level of the integration sphere are set to proper values.
• Then, the images from CCD1 to CCD5 are collected across different spectral bands and their average DN values are $\mathrm{D}{\mathrm{N}}_{1\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$.
• Next, the secondary mirror is covered with a black cloth, with the camera parameters and the energy level of the integration sphere unchanged. The images are captured again. The average DN values of the images from each CCD across different spectral bands are $\mathrm{D}{\mathrm{N}}_{2\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$. The configurations with and without the secondary mirror blocked during the test are shown in Figure 11.
• Subsequently, the integration sphere is turned off while keeping the camera parameters unchanged, and the dark field images are obtained. The average DN values of the images from each CCD across different spectral bands are $\mathrm{D}{\mathrm{N}}_{3\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$.
• Afterwards, the cover from the secondary mirror is removed and the Lyot stop is covered with a black cloth. The integration sphere is turned on while keeping the camera parameters and integration sphere energy level unaltered. From the image stored, the average DN values of the images from each CCD across different spectral bands are $\mathrm{D}{\mathrm{N}}_{4\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$.
• In the following step, the brightness of the integration sphere is increased to L1 (approximately six times brighter). The average DN values of the images stored from each CCD across different spectral bands are $\mathrm{D}{\mathrm{N}}_{5\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$.
• After shifting the camera parameters to night mode (10× reduced scan frequency, with the maximum integration level for each spectral band and 16× gain), the average DN values of the obtained images from each CCD across different spectral bands are $\mathrm{D}{\mathrm{N}}_{6\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$.
• Finally, with the integration sphere turned off, dark-field images are taken. The average DN values of these dark-field images from each CCD across different spectral bands are $\mathrm{D}{\mathrm{N}}_{7\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$.

Based on the data above, the calculation formula for ${\mathrm{V}}_2$ is as follows:

$${\mathrm{V}}_2=(\mathrm{D}{\mathrm{N}}_{2\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}-\mathrm{D}{\mathrm{N}}_{3\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}})/\mathrm{D}{\mathrm{N}}_{1\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$$

(6)

The veiling glare index ${\mathrm{V}}_2$ for CCD1 to CCD5 is shown in

Table 5. Veiling glare index ${\mathrm{V}}_2$ calculation results.

CCD No.	B1	B2	B3	B4	P
1	0.3349%	0.3386%	0.3305%	0.3309%	0.3368%
2	0.4091%	0.3700%	0.3701%	0.3686%	0.3865%
3	0.4807%	0.4230%	0.4074%	0.3866%	0.4055%
4	0.4406%	0.3911%	0.3835%	0.3808%	0.4026%
5	0.5106%	0.4097%	0.3902%	0.3793%	0.3876%

. The veiling glare index ${\mathrm{V}}_2$ with the secondary mirror covered is ≤0.52%. This indicates that the camera’s stray light suppression performance meets the usage requirements.

The veiling glare index ${\mathrm{V}}_3$ with the Lyot stop covered is calculated with the following formula:

$${\mathrm{V}}_3=(\mathrm{D}{\mathrm{N}}_{4\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}-\mathrm{D}{\mathrm{N}}_{3\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}})/\mathrm{D}{\mathrm{N}}_{1\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$$

(7)

The calculation formula for the veiling glare index ${\mathrm{V}}_4$ with the Lyot stop covered and the irradiance increased (approximately six times) is as follows:

$${\mathrm{V}}_4=(\mathrm{D}{\mathrm{N}}_{5\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}-\mathrm{D}{\mathrm{N}}_{3\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}})/\mathrm{D}{\mathrm{N}}_{1\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$$

(8)

The veiling glare index ${\mathrm{V}}_3$ and ${\mathrm{V}}_4$ are calculated according to Equations (7) and (8), and presented in

Table 6. The veiling glare index ${\mathrm{V}}_3$ calculated for different CCDs.

CCD No.	B1	B2	B3	B4	P
1	0.0877%	0.0141%	0.0062%	0.1236%	0.0424%
2	0.1443%	0.1043%	0.0926%	0.0724%	0.0221%
3	0.0369%	0.0684%	0.0309%	0.0295%	0.1352%
4	0.0361%	0.1087%	0.0802%	0.0317%	0.0232%
5	0.0127%	0.1305%	0.1295%	0.1331%	0.0553%

and

Table 7. The veiling glare index ${\mathrm{V}}_4$ calculated for different CCDs.

CCD No.	B1	B2	B3	B4	P
1	0.0673%	0.1262%	0.1371%	0.0666%	0.1139%
2	0.1918%	0.2081%	0.1947%	0.1484%	0.1131%
3	0.1066%	0.1824%	0.1783%	0.1505%	0.1637%
4	0.0896%	0.2085%	0.1684%	0.1299%	0.0778%
5	0.1846%	0.1926%	0.1613%	0.1088%	0.0993%

.

According to Table 6 and Table 7, the veiling glare index ${\mathrm{V}}_3$ measured with the Lyot stop covered under nearly typical operating conditions is ≤0.15%, while the veiling glare index ${\mathrm{V}}_4$ with the Lyot stop covered and irradiance increased (approximately six times) is ≤0.21%.

The differences between $\mathrm{D}{\mathrm{N}}_{6\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$ and $\mathrm{D}{\mathrm{N}}_{7\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$ are shown in

Table 8. The difference between $\mathrm{D}{\mathrm{N}}_{6\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}$ and $\mathrm{D}{\mathrm{N}}_{7\mathrm{CC}{\mathrm{D}}_{\mathrm{i}}{\mathrm{B}}_{\mathrm{i}}}.$.

CCD No.	B1	B2	B3	B4	P
1	1.7690	18.8162	13.2518	17.1797	2.4351
2	17.4308	6.7640	9.9600	31.0238	0.4854
3	17.4679	6.0951	9.4153	4.8390	4.8324
4	19.3198	39.9195	36.6368	10.0990	1.6273
5	0.6923	14.4641	6.9783	9.6559	0.2416

, which indicate the increase in DN value relative to the dark field under night mode with the Lyot stop covered and the use of the highest integration sphere brightness. It can be observed that in extreme conditions, the increase in DN is very small.

4.4. SPS Test

The transmission paths of stray light differ from those of imaging light. When the camera entrance is closed, imaging light is blocked and cannot form an image properly, but stray light may still reach the image plane through alternative paths. By closing the camera entrance and illuminating the rear of the camera with the light from an integration sphere, the detector’s response is entirely attributed to stray light entering through potential paths. This is known as the SPS test.

The SPS test system includes a 1.2 m-diameter integration sphere, a black cloth, the camera under test, and ground inspection equipment. The test setup is shown in Figure 12.

The test procedure is as follows: Firstly, the integration sphere is turned on, and its brightness is adjusted to a high value. The camera is positioned in front of the integration sphere, and the camera’s entrance is sealed to block the light. In these conditions, the integration sphere illuminates the back of the camera, and the camera’s response is recorded. Next, the integration sphere is turned off, and the camera’s response in dark field conditions is recorded. The camera is set to night mode (60× reduced speed). During the experiment, the average response value of the camera is observed, and the corresponding average response value under dark field conditions is also recorded. The detectors CCD1~5 are utilized to detect the incident light and record the response data. Since the data from each CCD are very similar, the response results of CCD1 are selected as a representative. The response difference of CCD1 between the integration sphere being turned on and the dark field condition is shown in Figure 13.

In Figure 13, as there is no active thermal control of the camera during this test, the slight increase in radiometric response after switching on the integration sphere may be related to a small temperature rise at the focal plane caused by radiation from the integration sphere. However, since the response difference curve does not show any significant abnormal variations, it can be concluded that no sneak path of stray light exists at the back of the camera. Meanwhile, the response difference is sometimes less than zero, which is caused by temperature drift.

Based on the above test methods, the stray light suppression performance of the GF04A satellite is evaluated. The results are as follows: (1) The ELE test shows that the stray light response ratio ${\mathrm{V}}_1$ outside the field of view is ≤0.12%. (2) The NPR test indicates that there are no ghost images in the sweeping direction of the camera’s field of view. (3) The KSR test shows that the veiling glare index ${\mathrm{V}}_2$ without the secondary mirror is ≤0.52%, the veiling glare index ${\mathrm{V}}_3$ for the blocked Lyot stop is ≤0.15%, and the veiling glare index ${\mathrm{V}}_4$ for the blocked Lyot stop with increased brightness (approximately six times) is ≤0.21%. (4) The SPS test shows that there are no obvious gaps in the camera’s rear cover. The above results demonstrate that the GF04A camera exhibits excellent stray light suppression capabilities.

5. On-Orbit Performance

The design target for the on-orbit absolute radiation accuracy of the GF04A satellite is better than 7%. As mentioned earlier, the stray light response of a space camera can affect the radiation accuracy. The prelaunch laboratory radiometric calibration carried out using an integration sphere gave a baseline for radiometric application, although it cannot reflect the stray light response of the system. However, during the on-orbit operation, the system is confronted with complex optical environment interference and is bound to be affected by stray light. Therefore, by comparing the on-orbit radiometric calibration results with those of the laboratory radiometric calibration before launch, the stray light elimination performance of the system can be reflected. According to the principle of error distribution, if the stray light coefficient of the system exceeds 5%, then under the condition of a radiometric calibration accuracy of 5%, the absolute radiometric accuracy of the system will be greater than 7% (the mean square sum of the stray light coefficient and the radiometric calibration accuracy).

Before launch, the GF04A satellite underwent laboratory absolute radiometric calibration experiments, with the results indicating an absolute calibration accuracy better than 5%. Then, after the GF04A was launched in 30 April 2022, an on-orbit calibration was taken using Baotou calibration site in 29 May 2022, based on synchronous ground radiation measurement. The error relative to the laboratory calibration coefficient, denoted as $\left|{\mathrm{L}}_{\mathrm{Measure}}-{\mathrm{L}}_{\mathrm{Onorbit}}\right|/{\mathrm{L}}_{\mathrm{Onorbit}}$, is obtained by subtracting the ground site calibration radiance from the observed radiance of the calibration site by GF04A camera, as shown in

Table 9. On-orbit calibration test results.

Spectral Band	${\mathbf{L}}_{\mathbf{Measure}}$ $\mathbf{W}/({\mathbf{m}}^2\cdot \mathbf{sr}\cdot \mathsf{\mu }\mathbf{m})$	${\mathbf{L}}_{\mathbf{Onorbit}}$ $\mathbf{W}/({\mathbf{m}}^2\cdot \mathbf{sr}\cdot \mathsf{\mu }\mathbf{m})$	Relative Error
B1	100.9941	101.4372	6.13%
B2	79.8411	85.0522	3.21%
B3	102.1205	98.9463	3.29%
B4	104.6978	108.2564	6.93%
P	80.1959	86.1673	0.44%

, where ${\mathrm{L}}_{\mathrm{Measure}}$ represents the observed radiance and ${\mathrm{L}}_{\mathrm{Onorbit}}$ represents the field calibration radiance.

As can be seen from Table 9, the relative error is less than 7%, indicating that the stray light suppression design is effective and the camera’s radiometric accuracy satisfy the design goal. A typical high-resolution image captured on orbit by the Jilin-1 GF04A satellite is shown in Figure 14.

6. Discussion

At present, in the field of high-resolution space camera development with diameters of several hundred millimeters or more, almost all stray light performance evaluations rely on simulation, and research work on physical testing is hardly seen. Therefore, comprehensive test approaches tailored to different types of stray light have been proposed in this paper that are distinct from conventional approaches such as PST test and VGI measurement using for small aperture systems. By evaluating the radiometric response induced by stray light, the proposed method enables a thorough assessment of the stray light suppression performance of remote sensing cameras using simple experimental setups. This approach avoids the need for expensive, dedicated, large-scale stray light test systems—which are often impractical or even impossible to construct—offering the advantages of low cost and high operational feasibility.

While traditional stray light tests can assess suppression levels, they suffer from limitations such as potential omissions due to angular scanning step sizes and the inability to trace the origin of stray light from the test results. In contrast, the joint test technique presented in this paper offers broader angular coverage, helping to prevent blind spots caused by special light paths. It allows for full-angle stray light magnitude measurement and the rapid identification of stray light sources when suppression structures fail. Additionally, the proposed test approach is derived from extensive experience in the development of spaceborne cameras. It demonstrates strong general applicability and is especially suitable for testing large-aperture cameras, with its cost-effectiveness increasing as the aperture size grows.

Nevertheless, the limitations of the proposed methods must also be acknowledged. First, there are measurement errors caused by differences between the laboratory environment and the on-orbit environment. Factors such as thermal variations, micro-vibrations, and long-term degradation in space may introduce additional stray light effects that are not fully captured during ground tests. Second, in terms of test coverage, the NPR test focuses on scanning-induced ghost images, which may overlook stray light effects from other incident angles. At the same time, the KSR test quantifies stray light propagating through key surfaces (e.g., the secondary mirror) by blocking them, but it may fail to account for complex multi-reflection scenarios. Finally, the proposed methods are specifically tailored to the optical design of the GF04A camera. Their applicability to other optical systems—particularly those with significantly different apertures or focal lengths—requires further investigation.

In future research, efforts will focus on further standardizing the proposed methods to enhance their general applicability. Real-time environmental data—such as solar angles and Earth albedo—will also be incorporated into stray light simulations to improve on-orbit prediction accuracy. During satellite development, new anti-reflection and absorptive materials for baffles and optical surfaces will be explored to further reduce BSDF and GCF. In parallel with this, automated test platforms will be developed to streamline the evaluation process, minimize human error, and improve repeatability. Finally, comparative studies will be conducted to assess the accuracy of different test methods, including comparisons between various laboratory approaches and on-orbit results, in order to identify common trends and system-specific challenges.

7. Conclusions

In order to minimize the impact of stray light, a dedicated suppression design is implemented for the Jilin-1 GF04A satellite remote sensing camera. We adopt an optimized combination of test methods—including ELE, NPR, KSR, and SPS tests—to thoroughly evaluate the camera’s stray light suppression capability. The test results show that the stray light response ratios obtained by the above test methods are consistently below 1%. These test results confirm the effectiveness of the GF04A camera’s optical design, laying a solid foundation for its excellent on-orbit performance.

After launch, the GF04A camera demonstrated outstanding imaging capabilities, with on-orbit calibration errors controlled within 7%. The on-orbit performance demonstrates that the camera achieves required radiometric accuracy, ensuring reliable data support for a wide range of remote sensing applications. Compared to traditional stray light evaluation techniques, the adopted test methods enable efficient pre-launch validation while avoiding the need for costly, large-scale test facilities. More importantly, the superior on-orbit performance of the GF04A camera fully validates the effectiveness of this integrated test approach and offers valuable insights for the development and test of future high-performance, large-aperture remote sensing cameras.