With the rapid expansion of a broad range of diverse systems, there is a growing tendency toward a role of the sensor as a means to obtain information. Especially image sensors, which can quickly capture a large variety of information across a wide measurement area, have lately attracted considerable attention. However, as conventional image processing technologies are based on the changes of gray scale, they confront some difficulty when attempting to distinguish overlapping objects in pixel image data. Their performance also declines against shadows cast by sunlight, low contrast period at dawn/dusk, which are typical situations for outdoor systems.

In order to resolve these problems, OMRON has developed and is now applying technology we call 'Silhouette Vision' as our proprietary approach to image processing of these kinds of problems. This technology uses two cameras to capture a stereo image, extracts characteristic points of an object from the image, and detects the object three-dimentionally in a scene. Moreover, we have developed various relational technologies, such as a camera control for stereo cameras, calibration techniques, model building algorithms using a 3-D recognition pursuit of the moving object, and outdoor environment resistance. Frequent field evaluation tests are revealing a level of detection accuracy never seen before.

OKAO Vision: Face sensing technology
Silhouette Vision: Stereo vision sensing technology
Pattern Vision: Character/symbol sensing technology