If only your physician could help you design your next machine-vision system.
Last month, as I was driving my Toyota Corolla back from an interview with Doug Wilson, president of PVI Systems (Niantic, CT, USA; www.pvisys.com), I became aware of a very sharp pain across the lefthand side of my chest. After pulling over to grab some fresh air and a cigarette, I got back in my car and drove home.
I told my boss. He informed me that a visit to the doctor might be in order. Otherwise, he said, who would fill the pages of the world’s favorite machine-vision/image-processing magazine? In this business, compliments get you everywhere, and so the next day I paid a visit to Dr. Beirmeiste, who performed an electrocardiogram and scheduled me for a cardiac stress test one week later.
Soon I was to find out why the test is called a “stress” test. After being hooked up to numerous electrodes, a nurse performed an electrocardio ultrasound on my heart. There, in living pseudocolor, I could see my ticker beating happily away on the 14-in. monitor of a Philips ultrasonic monitoring system. After making a few clever remarks about phased-array Doppler shifting techniques and how you could use the technology to detect and image submarines, I was informed that I was soon to mount a treadmill for 15 minutes and then the procedure would be repeated.
I was not amused. Exercise is not exactly myforte. Nevertheless, after running what seemed like 100 miles on the treadmill, my heart was imaged again. I watched as the nurse displayed my tricuspid, pulmonary, mitral, and aortic valves. These were certainly not designed by a human being. Nor did they resemble the simple valves in the Toyota that had carried me to the hospital on that rainy day.
As I lay on the gurney, I reflected on the subject of biomimetics-the concept of taking ideas from nature and implementing them in another technology such as engineering, design, or computer-based systems. Already, a number of subjects are being studied in universities worldwide. These include smart fabrics, food texture analysis, insect sense organs, and the structural makeup of animal bones. All of these projects have one idea in common-the process of natural evolution has led to the development of more sophisticated processes and designs than human beings can conceive, so why not copy them?
Image-processing and machine-vision designers have long sought to mimic these processes in the development of solid-state sensors, cameras, digitizers, CPUs, and neural-net-based imaging software. The reason many of the ideas fail is that they are seeking to emulate rather than copy the designs of a natural process. This evolutionary process has taken thousands of years and only now is mankind realizing its potential.
In his latest book,The Singularity Is Near: When Humans Transcend Biology, Ray Kurzweil argues that mankind is at the threshold of an epoch (the singularity being a reference to the theoretical limitlessness of exponential expansion) that will merge biology with genetics, nanotechnology, and robotics to create a species of higher intelligence. Although many people think that Kurzweil’s futuristic ideas are unimaginable, I am not one of them.
For example, building a machine-vision system that properly emulates the human visual system and cognitive process may be a difficult task, but it is made easier by developments in biology, nanotechnology, and computer science. And, thirty years from now, so-called “smart cameras” may become smart enough to emulate the human cognitive system.
After reflecting on this subject during my ultrasound, I pulled the electrodes from my chest and dreamed of a system that could “noninvasively” perform the procedure. But it was time to go home. And so, I wandered back to my Toyota Corolla.
Unfortunately, I had left the lights on, and the battery was dead. After another “stress test” with the local garage, the battery was replaced and I fired up the engine. Rather than learning from nature, some car makers may be better off learning from their customers and installing time-limit switches to stop this from happening. Product evolution is still a slow, slow process.