Embedded algorithm targets security
Many video, medical, and infrared reconnaissance images are recorded without controlled lighting conditions.
Many video, medical, and infrared reconnaissance images are recorded without controlled lighting conditions. As a result, features of interest within these images may occupy only a relatively narrow gray-scale range. To expand the contrast of features of interest so that they occupy a larger portion of the displayed gray-level range, contrast-enhancement techniques are often used.
One such technique—histogram equalization (HE)—produces a uniform histogram of the output image, so that an overall contrast is perceived. However, features of interest may require local enhancement. Using local-adaptive-contrast-equalization (LACE) software, the histogram of a local window centered at a given pixel can be used to determine the mapping of that pixel. In this way, the LACE algorithm performs individual contrast and brightness adjustments for each pixel.
With LACE, gain and offset values can also be calculated for each pixel individually. These values are based upon the statistics of the surrounding pixels, where nearby pixels have more influence than pixels farther away. Performing this algorithm in real time is computationally intensive and requires the use of a large number of multiply-accumulate circuits.
Consequently, implementing the algorithm in hardware proves more computationally efficient. At The Netherlands Organization for Applied Scientific Research (TNO; Delft, The Netherlands; www.tno.nl/instit/fel), a hardware version of this algorithm has been implemented in an ASIC, called VISTA01. During operation, the ASIC enhances the contrast of a video frame. This enhancement can be applied globally or defined by a user-definable area of interest. The contrast is evaluated in a gray-value histogram for each video frame. An application-specific MOVE-microprocessor reads the contrast from the histogram and calculates the parameters for the high-speed video path.
According to TNO, the algorithm can be tuned in software using an RS-232 data link to optimize performance when using specific cameras. For developers wishing to evaluate the algorithm, TNO has produced a dedicated PAL/NTSC-compatible demonstration board. On the board, analog-to-digital and digital-to-analog converters are used to convert the analog video to a digital format and then back to an analog signal for display purposes after processing.