Sensors cross the megapixel threshold
EETimes' Junko Yoshida describes a sensor battle that has surpassed the megapixel pursuit, beyond smartphones and tablets. In more complex and multi-faceted applications, the battle takes on a broader scope, defined by technical requirements demanded by end systems and engaging several players such as ON Semiconductor, CMOSIS, Teledyne DALSA, Sony, and Hamamatsu among others.
How often have I heard someone bragging that his camera has more megapixels than mine? Often enough that I can enjoy putting an end to that particular boast.
Nowadays, cameras are installed everywhere. Those eyes aren't there simply for watching you. They exist to help us—people, systems, and machines—to make decisions and take action. But for the machine vision to analyse what the cameras see, more megapixels aren't necessarily the answer.
Sure, pixels matter in smartphones and tablets, the biggest driving force behind the rapid growth of image sensors. And they matter when big design opportunities (typically represented by smartphones and tablets) are contested among leading CMOS image sensor vendors like Sony, Omnivision, and Samsung. Their battle is defined by the ever-escalating megapixel race, with Sony counting on 'selfies' and video calls to drive image sensor growth.
But the rest of the market—where image sensors are used in applications ranging from automotive and medical to sports analytics and food inspection—is a whole different story. That race is much more complex and multi-faceted. It engages many suppliers, including Aptina, ON Semiconductor, PhotonFocus, e2v, CMOSIS, Teledyne DALSA, Sony, and Hamamatsu. They compete on a variety of technical requirements demanded by end systems. Their criteria include frame rates, sensitivity, integration time, and light sources.
Megapixels, shmegapixels. Let's focus on innovations that go beyond the crude measure of counting tiny dots.
Consider the emerging automotive market. Many OEMs are racing to mount cameras everywhere on cars—for backup assistance, self-parking, lane departure warnings, lane changes, traffic sign recognition, night vision, and even driver drowsiness detection.
It's important to note that the functions of advance driver assistance systems (ADAS) are also progressing beyond a simple display of images on the centre-stack screen. Cameras are being designed to collect data for machines to see and prompt the car to respond. This shift towards intelligent analysis requires certain automotive image sensors to be less about megapixels and more about speed and sensitivity. This technology demands higher dynamic range sensors and near-infrared responses. Further, the auto industry expects image sensors of significantly better quality and reliability.
In January, the market research firm Yole Développement predicted that the CMOS image sensor market would rise from about $9 billion this year to about $13 billion by 2018. Eric Mounier, a senior analyst at Yole, said consumer applications would be "the growth driver over the next five years."
In particular, Yole predicted "a new wave of applications" in the automotive sector. "Along with the more traditional sectors of machine vision, security & surveillance and medical applications, these applications are likely to show strong growth in the mid and long-term."
Over the last two decades, the quality of CMOS image sensors has substantially improved. CMOS image sensors have also dramatically cut the cost of image sensor deployment in a system. Due to the relentless progress in cost and performance, CMOS image sensors have gone to places where no lens had gone before. We now see image sensors in places ranging from GoPro-like sports cameras and fingerprint scanners to currency verification machines and dental scans.
"The applications of image sensors are only limited by your imagination," Edwin Ringoot, strategic marketing manager for the image sensor business unit at ON Semiconductor, told EE Times last week.
The following slideshow depicts the ubiquity of image sensors, all of which have cleared the high technical hurdles imposed by specific applications.
|Related Articles||Editor's Choice|
|Related Articles||Editor's Choice|