March 14, 2012
Background
Infotech Inc. of Japan sold a night vision camera system manufactured in the U.S. to a Japanese electric company for surveillance in2008. The system was deployed at three nuclear power plant sites in Japan,
primarily for sea surveillance. However, even though the system used a cutting-edge laser camera, it could not perform efficiently in dense fog. To address this problem, Infotech Japan started to develop an image processing technology that could actively correct for this condition. Mr. Masahiro Kobayashi of Infotech Japan invented a unique algorithm for the processing now known as the “Detail Enhance Filter” (DEF). The patent application has beenregistered with the Japanese and the U.S. governments and is now published.
This image processing technology was developed to integrate not only with specialized night vision camera systems, but also with color camera systems used for surveillance and other applications such as Near Infrared (NIR) images. Infotech North America was incorporated to implement production in the U.S and is introducing an upgradeable commercial stand-alone video processing platform that can be integrated with many existing video camera systems for the global military, security and surveillance, medical, public safety, and vehicle mounted video markets.
Overview
DEF is a unique image improvement filter, with a broad range of applications in surveillance and data capture cameras. It works with existing video recording networks, as well as for real-time viewing of imagery as it is collected. The device will enhance the detail of subjects in any given image captured under unfavorable climatic and lighting conditions, including fog, snow, rain, low light, strong backlight,
sun glint, and many other circumstances where the image lacks sufficient viewable details in all or portions of a video frame.
Purpose
The image processing can be used for surveillance and security camera systems, underwater research, bio-medical images, surface inspections, and any other application that incorporates a camera system. The DEF algorithm takes advantage of the full available dynamic range of the image signal, in context of the surrounding image information, to enhance both surface or edge detail and to extract image information in dark or bright regions of the image. The robustness of the algorithm allows for an image with both overly bright and dark areas to be corrected simultaneously. Adaptive calculation for image sharpening and enhancement is automated and requires no additional user configuration of the parameters that adjust for varying image conditions such as fog or darkness. The operator hasthe option to control the blend of the original and processed images to regulate the level of enhancement required to produce a clear image.
Features
- Automatic Operation - No need to set parameters, but intensity level can be changed
- Real time Processing - Delay time within 1 frame (1/60 second) when using HD video
- External Control – Various serial or ethernet communications configurations
- Four channel SD format (BNC) or HD format (HDMI) video daughter cards in development
- Works with live or recorded video and still imagery
- DEF is a Plug and Play addition to existing video collection and monitoring systems
Algorithm
(U.S. Publication Number: US-2011-0273748-A1; Publication Date: 10 November 2011) Video picture quality processing is performed by means of histogram averaging, which is one means of improving contrast in an image. However, it takes time to calculate the histogram for an entire image with a high numbers of pixels, and requiring large quantities of hardware processing resources. Image processing for each frame of a video, or for a static image, is comprised of a loading device, which loads pixel-unit image data from an image and projecting a photographic subject, and a histogram generation mechanism, which generates a histogram of the brightness after subsampling and breaking down the loaded image data into specific color spaces. The generated image is read out in patterns designated for each color, and the brightness is set for pixels at specified locations in the aforementioned patterns, based on the average histogram while excluding the pixels at specified locations in these patterns.
Production Schedule 2012
Demonstration units are being field tested by selected clients while production is being implemented. It is anticipated that approximately 1000 units will be manufactured in 2012 with delivery slated to begin in July. Designs for the next generation of DEF systems, which will offer additional options, such as customer specified video I/O interfaces, will be also available.
Contact Bill Bernard at 443.786.4471 and Email Contact for further information.
