How spectral sensing solves true-to-life camera color constancy in challenging lighting conditions
New white paper shows how multi-channel spectral sensors identify light sources for accurate automatic white balancing (AWB) and true-to-life imaging.
Fierce competition drives manufacturers of smartphones, smart glasses and action cameras to regularly introduce hardware and software innovations aimed at improving camera image quality. Recently, a new front in this battle has opened, sparked by concerns over the unreliable operation of cameras’ automatic white balancing algorithms in certain circumstances.
A 2024 research paper published by the International Journal of Computer Vision showed how spectral analysis of lighting conditions could solve the automatic white balancing problem. ams OSRAM has validated the technique, showing how manufacturers can use a chip-scale sensor to capture the spectral signature of ambient light for accurate image white balancing.
The problem with today’s automatic white balancing algorithms, the principle underlying the use of spectral sensing to identify light sources, and the results of validation testing are all described in a new white paper about spectral sensing for accurate automatic white balancing. Download the white paper about spectral sensing for accurate automatic white balancing, and learn how the latest ams OSRAM multi-channel spectral sensors give camera manufacturers a new edge in the competition to attract consumers’ dollars.
Ambient light sensors evolve beyond RGB to enable true-to-life camera color accuracy with XYZ spectral sensing
At ams OSRAM, we have for many years been leading the development of technology for measuring ambient light in small chip-scale form factors which fit easily into the circuitry in digital and smartphone cameras. Starting with the earliest two-channel and RGB color sensors, our products have evolved to offer more accurate measurement of color, and more sophisticated ability to distinguish the sources of ambient light. This has made us a highly trusted partner to mobile device manufacturers, supplying light-sensing technology products to them in high volume.
The latest breakthrough for enhancing the operation of automatic white balancing has come with the introduction of a new multi-channel spectral sensor, which combines XYZ sensing with spectral analysis in a device optimized for use in smartphones and other mobile or portable products.
So what is the problem with automatic white balancing that this spectral sensor solves? And why is accurate light source identification so helpful in improving camera performance?
The problem of a single-color background
In most scenes, digital cameras can achieve the correct white balance using only data derived from the main RGB image sensor. But we also know that cameras’ automatic white balancing encounter problems with rendering color accurately in scenes with a strong single-color background, especially if it is blue, brown, pink or purple. These monotone backgrounds trick the algorithms for converting the color that an image sensor ‘sees’ in distorting lighting conditions (such as outdoors at dawn or dusk, or fluorescent indoor lighting) to the color that the human eye recognizes as natural.
The algorithms try to infer the color profile of the light illuminating the scene from the image of objects in the scene. This inference technique fails when certain background colors are dominant. It turns out that the solution is to identify the signature of the light source so that it can be accurately classified. When the light source is known, the appropriate color correction can be applied in software to achieve excellent white balancing.
Now our new white paper shows that a multi-channel spectral sensor can detect the distinctive spectral signatures of sunlight, LEDs, fluorescent lighting, and all other common types of light source, and so provide the camera’s algorithm with the information that it needs to apply the appropriate color correction.
The result: ideal color constancy in all lighting conditions. A camera equipped with an ams OSRAM spectral sensor can produce natural images, for instance accurately rendering skin tone in pictures of people’s faces, a type of scene in which color constancy is particularly noticeable. And because the sensor detects the spectral signature of the light source, rather than interpreting the colors of objects in the scene, it works in all lighting conditions and all types of scenes.
Simulations validate spectral sensor performance
The new white paper has given a solid statistical foundation to the idea that a spectral sensor can identify light sources correctly. We have performed a detailed series of laboratory tests which simulate the output from a typical smartphone camera’s image processing system when attempting to identify a scene’s light source using either the RGB image sensor or an 8-channel spectral sensor.
The difference is stark: when using the spectral sensor, the error rate in detection of the light source falls to close to zero. This shows that manufacturers can use this new technique to achieve accurate automatic white balancing across all lighting conditions and so eliminate the color constancy problem which today’s cameras suffer from.
Market research has shown that perceived image quality is one of the factors that consumers care most about when choosing a mobile device. By integrating a spectral sensor into the image processing system for accurate white balancing, a camera can produce images that look natural, with true-to-life color in every scene.
Details about the tests conducted and the results gained from them can be found in the white paper. You can download the paper now. And if you want to explore the application of spectral sensing further, contact ams OSRAM to learn more about the use of spectral sensing for accurate image white balancing.
Spectral sensing for AWB white paper
Sign up now!
フォーム読み込み中...
Your data is used only for internal purposes related to the registration process.
All fields marked with an asterisk(*) must be filled out.