Figure 4(b) shows the effect of the proposed method on
an image with average brightness. The image 4(a) is the
original while 4(b) was enhanced using our algorithm.
For this image the dynamic range of the picture
increases. Since most of the information is in the mid
luminance range, making dark pixels darker increases the
contrast and provides a better visual experience. In
another example, Figure 4(d) shows the effect of the
algorithm on a dark image. In this case most of the
information lies in the dark region and therefore making
dark pixels darker would not help. . It can be seen from
the enhanced image, that the dark pixels are made lighter
in this case so that details are enhanced. At the same time
the people’s faces and other bright pixels are not
modified and so the algorithm gives a localized contrast
enhancement effect. In many video applications, the
pictures may have a letterbox which is usually dark. If
the letterbox regions are used for HE processing, they
will skew the histogram and the final output will be suboptimal.
The letterbox region will be processed by HE
and may become lighter, which is not desirable. This
problem is solved by doing automatic letterbox and
pillarbox identification at the front-end of the ACE block
and extracting only the active window of the frame for
processing by the algorithm. The proposed method was implemented on an ASIC and tested with real video. For
processing video frame storage was avoided by using the
statistics collected from the previous frame to process the
subsequent frame.