In daily life, have you ever encountered such a situation: when shooting against the light, the entire image looks like it is covered with a “veil”, the image contrast decreases, and circular light spots appear (see the figure below). These phenomena are all caused by flare! Flare reduces image contrast, obscures details, and may introduce interfering information, negatively affecting image quality and analysis results. Therefore, flare testing is a crucial part of digital camera image quality testing. This article will introduce the principles behind the flare phenomenon in digital cameras!

I. What is Flare?

Flare refers to non-imaging light that reaches the image plane in an imaging system.

As shown in the figure, imaging light rays (solid blue lines) undergo unexpected reflections inside the lens, generating non-imaging light rays (dashed blue lines). These non-imaging light rays that do not propagate along the intended optical path are collectively referred to as flare. When flare reaches the image plane, it interferes with normal imaging.

II. Common Forms of Flare

The impact of flare on imaging manifests in various forms:

For example: in Figure 3, under strong sunlight, the buildings appear to be “covered with a veil”, exhibiting a veiling glare phenomenon with a large-area reduction in image contrast and multiple blurred circular light spots; in Figure 4, petal-shaped red flares bloom around the sun; in Figure 5, a string of colored light spots (ghost reflections) trails below the sun; and in Figure 6, radial lines resembling starbursts emanate from the sun (diffraction spikes).

III. How is Flare Generated?
On a macroscopic level, the flare phenomenon is caused by unexpected reflection, scattering, or diffraction of light within the imaging system. Specifically, it can be attributed to the following categories:

• Flare caused by single or multiple reflections between lens elements, between lens elements and structural components (such as the lens barrel, edges of the aperture stop, etc.), between lens elements and camera structural components (such as the lens mount, brackets, dark chambers, etc.), and between lens elements and the image sensor;
  
• Flare caused by scattering (due to scratches or dirt on lens surfaces, inconsistencies in lens materials, etc.);
  
• Flare caused by the diffraction of light by pixels on the solid-state image sensor.
  

For example: the veiling glare in Figure 3 is caused by the scattering of non-imaging light inside the lens; the petal-shaped flares in Figure 4 are formed by diffraction spots generated by the sensor's microlens array and reflected by the filter; the string of circular light spots in Figure 5, commonly known as ghosting, is caused by multiple reflections of light between lens elements; and the starburst effect of the sun in Figure 6 is often caused by diffraction formed by the aperture stop of the lens.

IV. Flare Testing Standards
Three common standards for camera flare testing:
1. ISO 18844:2017 Flare Testing Method
The standard for measuring camera flare in the field of photography is ISO 18844:2017. Its core approach involves setting up multiple circular black areas with low reflectance/transmittance (i.e., a diagonal array of high-density light traps (black dots), see Figure 7) on a uniformly illuminated or uniformly luminous white background. By calculating the luminance ratio of the black areas to the white background in the image, known as the flare index F, the severity of the flare is quantitatively evaluated. The smaller the F value, the less impact the flare has on image quality.

Figure 7 Diagonal array of high-density light traps (black dots)

Calculation formula for flare index F:
$$F = \frac{S_{black}}{S_{white}} \times 100\%$$

Figure 8 ROI area for F value

2. QC/T 1128-2019 Testing Method
QC/T 1128-2019 specifies the testing methods and evaluation metrics for the flare performance of automotive cameras. In QC/T 1128, the device under test (DUT) must capture images of a point light source in a dark environment. The camera's ability to suppress flare is evaluated based on the proportion of glare and ghosting in the total image area as defined in the standard.
Glare Metrics:
The area of the light spot generated by the lens under test light illumination shall not be greater than 25% of the display area.
Glare Testing Method:
a) Without the light source, the luminance of the fully black test chart shall be less than $2\text{cd/m}^2$;
b) The light source shall be (5～10) $\text{Mcd/m}^2$, occupying (30～34) $\text{arcmin}$ in the field of view;
c) Ensure the light source is within the field of view of the DUT, and adjust the incident angle of the light source to maximize the area of the light spot in the image;
d) Using video analysis equipment, calculate the area occupied by regions with a signal intensity greater than 50% of the maximum intensity in the image, and calculate the ratio of this area to the entire image area.

Ghosting Metrics:
a) If the ratio of the peak luminance of the ghost image to the original image is greater than 50%, the area ratio of the ghost image to the field of view shall be less than 1%;
b) If the ratio of the peak luminance of the ghost image to the original image is greater than 30% and not greater than 50%, the area ratio of the ghost image to the field of view shall not be greater than 8%;
c) If the ratio of the peak luminance of the ghost image to the original image is not greater than 30%, there is no requirement for the area ratio of the ghost image to the field of view.
Ghosting Testing Method:
a) The point light source has a power of 1W, a color temperature of 6000K, a luminous angle of (110+10)° (half intensity), and an illuminance of (220+22)lx. The DUT is tested in a dark chamber in operating mode B1.
b) The distance from the DUT lens to the light source is (400±10)mm. The lens is fixed on a rotating stage facing the light source, and the image is confirmed to be normal;
c) Rotate the DUT clockwise in the horizontal direction, with the center of rotation being the geometric center of the lens. The images after each rotation are separated by 10% of the horizontal field of view, until the light source exceeds the 40% area of the horizontal field of view, and save the images for each field of view;
d) Reposition the DUT to face the light source, and rotate it counterclockwise in the horizontal direction. The images are separated by 10% of the horizontal field of view until the light source exceeds the 40% area of the horizontal field of view, and save the images for each field of view;
e) Statistically analyze and calculate the area ratio and luminance ratio of the ghost images in the images of each field of view.

3. IEEE 2020-2024 Testing Method
The IEEE 2020-2024 standard is an important specification for automotive image quality evaluation. Its defined flare testing includes two testing methods: Flare A and Flare B:
(1) Flare A Test:
Based on the standard dot matrix chart of ISO 18844:2017, additional test points are added on the horizontal and vertical axes to cover a larger field of view and capture flare in specific directions. The severity of the flare is quantitatively evaluated by calculating the flare index F value;

IEEE 2020-2024 Standard Flare A Test Chart

(2) Flare B Test:
By rotating the point light source or the device under test (DUT), it simulates a strong light source incident on the lens from multiple angles such as horizontal, vertical, and diagonal. After capturing a series of images, the flare attenuation of the images is analyzed, and metrics such as Average Flare Attenuation and Worst Flare Attenuation are calculated to quantify the camera's flare performance.

Fixed DUT, rotating light source	Fixed light source, rotating DUT

Example test image

Average Flare Attenuation
The arithmetic mean is taken for the $FlareAverage_{dB \theta，\phi}$ values (i.e., the flare attenuation value of each pixel) of all pixels in the entire image. The calculation formula is as follows:
$$FlareAverage_{dB \theta，\phi}=avg(Flare_{dB \theta，\phi}(x,y))$$ $FlareAverage_{dB }$ is the average flare intensity, in decibels (dB); $\theta$ is the azimuth angle of the light source; ${\phi}$ is the field angle of the light source; x is the image coordinate x; y is the image coordinate y; $Flare_{dB }$ is the flare attenuation, in decibels (dB).

Worst Flare Attenuation
The data with the maximum flare intensity in the image is taken. The calculation formula is:
$$FlareWorst_{dB \theta，\phi}=\max_{x,y}(Flare_{dB\theta,\phi}(x, y))$$ Where: $FlareWorst_{dB}$ is the worst flare attenuation, in decibels (dB); $\theta$ is the azimuth angle of the light source; ${\phi}$ is the field angle of the light source; x is the image coordinate x; y is the image coordinate y; $Flare_{dB }$ is the flare attenuation, in decibels (dB).