Spectral sensors for automatic white balancing (AWB) enhance color consistency

Today’s smartphone cameras can be fooled by the colors in some scenes. Multi-channel ams OSRAM color sensors ensure that images are color-perfect in every shot.

AWB (automatic white balancing) is one of the essential techniques deployed in smartphone cameras to produce color consistency: what you should see in an image on the phone’s display screen should be what the scene looks like in real life. Now smartphone manufacturers are discovering that spectral sensing can solve a problem with automatic white balancing that has emerged in certain hard-to-handle scenes.
 

Why color consistency really matters to smartphone users

It is only as recently as 2007, the year when Apple introduced its iPhone® device, that it first became normal for consumers to carry a digital camera everywhere they went. The camera quickly became a ‘killer app’ for smartphones. Then came image-based apps such as Instagram and Tiktok, and photographically speaking, the rest is history.

Enhanced by the awesome algorithmic power of the processor in today’s phones, the camera can produce near professional-grade pictures in many indoor and outdoor scenes. But the pin-sharp resolution and high color sensitivity of the camera only serve to accentuate flaws in the camera’s operation. One such flaw which is becoming increasingly apparent to phone users is a failure to maintain color consistency in certain types of scenes.

When the white point reference is wrong

Phone users are particularly sensitive to inaccurate color in pictures of people, because color inconsistency is readily apparent in skin tone and the color of clothing.

In general, what the user wants to see in an image is ‘true to life’ color: the objects in the image should appear natural to the human eye when rendered on a good-quality display. But some lighting conditions distort the appearance of colors, making them look unnatural. At dusk, for instance, white colors become unnaturally golden. Under typical office LED lighting, the same colors will look more blue than normal.

So smartphone cameras apply a color correction process called automatic white balancing, or AWB, which is meant to make the average color in the scene appear neutral, as it would under typical natural light (ie daylight). To perform AWB, the camera has to establish a reference white point, often by finding a white object in the scene, or by using a technique such as the ‘grey world algorithm’. Recently techniques based on machine learning have begun to be used for AWB.

A typical case in which a smartphone camera (left-hand image) is fooled by a solid colored background. The image on the right side, taken by a camera with automatic white balancing enabled by a spectral sensor, has the correct color balance, shown by the natural looking appearance of the flowers.

But these techniques that the camera uses to estimate the color cast can be fooled in certain types of scenes, leading its AWB function to apply the wrong color correction shift, and so to produce unnatural looking colors. This effect is particularly noticeable in scenes shot against a monotone blue, brown, purple, or pink background.

 

Detecting a unique light signature

Now, advanced spectral sensor hardware from ams OSRAM enables smartphone cameras to improve white balancing by providing valuable additional information about the lighting conditions.

Every light source has its own ‘spectral signature’: the spectral power distribution (SPD) curve for a cool white LED lamp, for instance, looks quite different from that of a warm white LED or of bright sunlight.

The SPD curves of different light sources have sharply contrasting signatures

Spectral sensors can detect this spectral signature, enabling the camera to identify the light source, and then apply the appropriate white balancing. Spectral sensors can also provide a very accurate measurement of the color value of the scene being photographed.

If the camera knows the light source, it can apply color correction tuned for that light source. It does not need a white point reference, or in fact any information drawn from the objects in the scene at all. All it needs is the spectral signature which identifies the light source or its corresponding color coordinates. The camera then uses pre-programmed information about the characteristics of that light source to apply the appropriate correction shift to all the colors in the scene.

In addition to white balancing, spectral information about the light source helps the camera to compute a more accurate color conversion matrix (CCM), as well as perform better lens shading correction, which are essential parts of the image processing pipeline.

Spectral measurements of this kind have long been available to specialists and professionals: they are produced by a spectrometer, a type of laboratory equipment which costs many thousands of dollars. ams OSRAM has developed spectral sensing technology which can be implemented at chip scale, making it physically possible, and economically affordable, for manufacturers to integrate a spectral sensor alongside the image sensor in a smartphone’s camera.

By slicing the spectrum of light into as many as eight channels, the camera’s spectral sensor can detect the distinctive blue peak of a cool-white LED at around 450nm, or the characteristic red peak of a warm-white LED at around 650nm, enabling the camera to identify the unique signature of any commonly used light source, and so to apply the appropriate color correction values.

The result? Color-perfect images every time, even in scenes which fool today’s cameras’ estimation of the neutral white point.

There is huge reward on offer for smartphone manufacturers which solve the AWB problem: DXOMARK has published results from tests of smartphones ranked by color constancy and camera performance: it has found that consumers strongly prefer smartphone models which offer superior image quality and color consistency.

This is what the spectral sensing technology from ams OSRAM offers, both for the camera’s AWB function, and equally for the display: here too, identification of the light source enables accurate color management, so that the true to life colors captured by the camera are also rendered true to life by the display, producing what is known as ‘glass-to-glass’ color consistency.

Start a conversation about spectral sensing with ams OSRAM today, and find out how true color consistency elevates the value of next-generation smartphone cameras.

True-to-life color displays with spectral color sensing behind OLED

Form loading...

\n