Seeing Value in CMOS
James He is chief operating officer at OmniVision Technologies. In this Visions, he talks with Security Magazine about the CMOS role in security.
Security: First if all, what types of image sensors are commonly used in security applications? Can you give an overview of what, in your opinion, the relative strengths of different imaging sensor types are?
He: There are two main types of image sensors commonly used in the security market – CCD (Charge Coupled Display) and CMOS (Complementary Metal Oxide Semiconductor).
For many years CCD had the larger market share and was the “technology of choice” as it was the only really viable solution. CCD offered high-performance imaging compared to previous generations of CMOS sensors, but demanded that the product be sufficiently large to house multiple system components. CCD solutions often comprised expensive multi-chip systems, were also complex, and based on a highly specialized process. As a result, users had to contend with the resolution, range and temperature limitations of CCD sensors - issues that are important in security applications.
Over the past several years, CMOS technology has evolved and made headway in entry-level commercial security applications. This advance is a result of lower power requirements along with significant improvements in performance and sensitivity. As pixel sizes have become progressively smaller, CMOS-based solutions became more cost-effective. CMOS image sensors integrate circuitry on a single chip in a compact design and enable a smaller camera unit than those found in competing CCD systems. CMOS image sensors typically use just 1/10th of the power used by CCD devices, a factor which, when multiplied out over today’s large security networks, equates to substantial financial savings for the user. CMOS image sensors are less expensive to manufacture and are typically sold at a lower cost, again lowering overall system cost, meaning further cost savings to manufacturers.
Security: How do you expect to see CMOS image sensors being employed in the commercial and domestic security markets, especially considering the different requirements of low-end, medium and high-end systems?
He: We expect to see cameras using CMOS sensors become firmly established as the technology of choice for security applications in terms of market share. We are already seeing the mass market deployment of CMOS sensors in the Asian OEM and ODM markets, where we have many customers in the Webcam and security camera sectors.
The accepted path for technology is that it moves from high-end applications down to mass production as it becomes more cost-effective. The commercial security camera domain does not follow this trend, as we are seeing CMOS sensors, currently at the heart of the majority of consumer cameras, quickly gaining traction in high-end commercial applications. The gap in image quality between CCD and CMOS image sensors has narrowed greatly – so much so that a new wave of CMOS-based cameras is making its way into production and into mid- and high-level applications.
We feel that the industry is now starting to see that CMOS imaging solutions offer a combination of lower cost, rapidly advancing performance, less power drain, dramatic advances in light sensitivity, and a critical size advantage. CMOS sensors increasingly address the high-volume, space-constrained needs of commercial surveillance, where performance is the end goal.
The new generation of CMOS-based systems owes its increasing popularity to the complexity of commercial surveillance equipment entering the marketplace. The focus of this equipment is on size, low power, the ability to support intelligent, anywhere/anytime security, and having the video output available globally. The fundamental issues at work to provide this complex equipment involve the number of chips, the total integration of necessary features and the best resolution across often widely varying environmental conditions.
Security: What are the key considerations, in your opinion, when designing a camera or video security system?
He: One key element in the design of a camera-based security system is the actual image sensor used. Different applications have, of course, different demands, but also some common requirements. The performance of any particular sensor needs to be sufficient to meet or exceed the expectations of the industry. Image quality, performance in varying light conditions and at varying temperatures, are all important elements in sensor choice, allied to the overall bill of materials, ease of integration into a camera design or security system, and end user installation, operation and maintenance.
Connectivity is an extremely important design consideration for the security market. Looking at connecting technologies, we see wireless connectivity and secure Internet playing a larger role in security applications.
IP cameras can be networked either wirelessly or via cable for on-site or remote monitoring via the Web, both for commercial and domestic security applications. IP cameras give the benefit of higher resolution (no longer limited to NTSC/PAL standards) and broader accessibility (from point-to-point CCTV to an Internet accessible network).
IP-based security will allow for the adoption of much more sophisticated software applications and also facilitate the adoption of digital output CMOS sensors with higher resolution levels and additional features, such as enhanced night vision.
Security: Looking to the future, which areas do you feel will be most important as the technological demands of the security market increase?
He: There are three main factors that will become critical to address the demands of the security market: intelligence, integration and innovation.
Intelligence: Commercial security cameras have become smarter. As a result, it is becoming increasingly necessary to intelligently analyze collected data in real time. CMOS technology first enables intelligence by providing enhanced digital details required by motion detection, face recognition and other image analysis applications. CMOS enables a user to access a segment of the image rather than looking at the entire image captured. Each CMOS active pixel sensor cell can be addressed and read individually using a simple addressing technique. Competitive technologies require the entire contents to be read out, which in turn negatively affects performance.
Smart CMOS-based cameras make real-time decisions that dramatically enhance both performance and quality of image. The result is a higher level of security protection and the ability to avoid the errors inherent in consistent human monitoring of surveillance video.
Integration: Further integration is an area we feel will be critical in future applications. With CMOS sensors, for example, the ability to receive and convert light to digital output is embedded on the sensor. Several features reside in the one-chip solution, including analog-to-digital conversion, electronic shutter, trans-impedance amplifiers, and sample-and-hold circuitry, as well as application related features such as back light compensation and lens distortion correction. When the majority of high-end surveillance cameras were analog, the need for multiple chips wasn’t an issue. Today as a rapidly growing number of surveillance cameras are digital, it is. The one chip solution allows for smaller cameras without a performance penalty.
Innovation: As CMOS technology continues to make headway into high-performance applications, there are a number of advances. For instance, the number of transistors in each pixel is rising. Today, up to seven transistors are resident in each pixel. Each pixel in effect becomes a camera - and photo quality is dramatically enhanced since auto exposure is performed on the complete array. CMOS is the only technology that can deliver a camera on a chip where the only ancillary elements are an oscillator and a power supply.
Innovations such as color HDR (high dynamic range) enable CMOS sensors to function much like the human eye, adapting to rapidly changing light conditions.