New Image Sensors Find Use in Brain Imaging, Security Cameras, and Even Selfies

January 21, 2021 by Nicholas St. John

A new crop of image sensors shows how device makers have been thinking out of the box with on-chip hardware "re-mosaic," curved image sensors, and more.

If CES 2021 proved one thing, it's that seemingly everything has a camera on it in the electronics industry. And these high-performance image sensors are making an appearance in all areas of the market, from medical applications and security systems to smartphones featuring "the world’s best selfie."

Recent acquisitions have demonstrated an interest in image sensor development, too. For instance, Teledyne Technologies recently acquired and FLIR Systems—a known specialist in thermal imaging cameras, components, and imaging sensors—for $8 billion.

What are some recent advancements at the component level that are illustrating innovation in image sensing technology? 


An Image Sensor for the Perfect Selfie?

OmniVision, a leading manufacturer of advanced imaging solutions, recently released an image sensor for smartphones, the OV32B.

One of the most impressive aspects of this sensor is its size in relation to performance: it has a compact ⅓” optical format, which leads to an approximate diagonal size of the image sensor being only 5.5 mm, but still giving 32 megapixel (MP) resolution with a pixel size of 0.7 microns. This is a helpful feature for smartphone designers since they have limited space to work with.

The sensor also supports 2- and 3-exposure HDR (high-dynamic-range) timing for up to 8 MP video modes and still previews. OmniVision says this creates a more visually-impactful picture by adding more contrast and color to the end product.


Functional block diagram of the OV32B40

Functional block diagram of the OV32B40. Image used courtesy of OmniVision (PDF)

For color optimization, the sensor offers a 4-cell color filter array and an on-chip hardware re-mosaic that provides a 32 MP Bayer output. This means that the image sensor is arranged in a way that neighboring pixels each have a different color filter on them—either red, green, or blue.

The sensor then uses near-pixel binning in low-light conditions, increasing the pixel size to 1.4 microns and increasing pixel sensitivity to four times the original value. OmniVision touts that this feature allows the image sensor to capture high-quality selfies, even at night.  


Curved Image Sensors

Imaging a wide area is a challenge because naturally flat optics tend to curve the surface of the image. For this reason, extra hardware must be added to correct these inaccuracies. There is, however, another option that utilizes curved image sensors, which naturally correct the issues associated with flat optics.

This year, French startup Curve-One announced its own curved image sensor—the first of its kind to be commercially viable, according to the press release. While Curve-One is not the first to make these curved image sensors—Sony created a pair in 2014 and CEA-Leti is creating a 20 MP prototype—the ability to be commercially manufactured is a notable capability.


The curved image sensor

The curved image sensor. Image used courtesy of Curve-One

The company claims its optics can reduce the hardware amount to one-third of the requirement for flat optic sensors. Additionally, curved sensors reduce the number of errors and misalignments in the complete system, which may provide more consistent results.

Curve-One discusses how its manufacturing method is bio-inspired, mimicking the eye’s retina. These curved image sensors are being used in the Meso-Cortex project, a project that requires high-resolution images of the human brain.


Shedding Light on Security

In addition to OmniVision's selfie-centered image sensors, the company is also honing in on security systems as well. Another sensor, the OSC04C10, features low power consumption, selective conversion gain, staggered HDR, and exceptional low-light capability, according to the press release.

The sensor is also said to measure near infra-red (NIR) light (850 to 940 nm); since this kind of light is not visible to the naked eye, it enhances images in low-light conditions. OmniVision says the OSC04C10 is able to capture NIR light without compromising the color precision of the resultant image, too. 



Functional block diagram of the OSC04C10. Image used courtesy of OmniVision (PDF)

The sensor has 2.0 micron pixels with 4 MP resolution, specifically a 2688 x 1520 resolution within a 16:9 aspect ratio. It is said to provide this high resolution at 60 fps (frames per second). In order to accommodate security applications where the system is continuously on, there is an ultra power mode for this sensor that consumes 98.9% less power than when the sensor is in its normal operating mode.



The new image sensors listed here are just a few of many that may shape future imaging systems. What experience do you have with designing such systems? Share your thoughts in the comments below.