Exploring High Dynamic Range Imaging: §3.5 Adaptive Gain Control Local Operator with Edge Detection

Module by: Tianhe Yang, Taylor Johnson, Sarah McGee, Robert Ortman

To lessen the blurring effect on the adaptive gain control operator, we modified Pattanaik’s algorithm so that edges can be detected. In the case where a neighbor pixel is “too different” in intensity from the pixel being mapped, a counter is incremented and counts the maximum number of pixels that can be “too different” before the inspected pixel is not mapped at all using the averaged obtained from its neighbors. In this case, the old luminance value is simply assigned to this pixel, which is mapped down to a lower color depth linearly by dividing by the brightest value in the image. In the following examples, the “max difference” parameter refers to the maximum contrast ratio between a neighboring pixel and the pixel being mapped, above which a counter is incremented. The “number different” parameter refers to the maximum value of said counter, above which the local average can be ignored when mapping a pixel. As expected, increasing the “max difference” parameter increases the threshold at which edges are preserved, thereby producing a blurrier image. A small value for “max difference” keeps finer details intact as slighter changes would prompt the mapping algorithm to ignore local values. Increasing the “number different” parameter has a similar effect as increasing the “max difference” parameter; as more values must reach the threshold before the local average is ignored, the local average is used more often to map image pixels, causing the image to become more homogenous. Blurring the image by increasing the “number different” parameter, however, seems to keep the details in harsher edges more intact compared to blurring with the “max difference” parameter. This should make sense (pixels at the edge of a harsh transition encounter a greater number of neighboring pixels with very different luminance values).

Figure 1: Threshold = 5, Radius = 2, Max Difference = 1.1, Number Different = 1

Figure 2: Threshold = 5, Radius = 2, Max Difference = 1.35, Number Different = 1

Figure 3: Threshold = 5, Radius = 2, Max Difference = 1.75, Number Different = 1

Figure 4: Threshold = 5, Radius = 2, Max Difference = 1.1, Number Different = 3

Figure 5: Threshold = 5, Radius = 2, Max Difference = 1.1, Number Different = 10

Comments, questions, feedback, criticisms?

Send feedback

• E-mail the authors of the module, Exploring High Dynamic Range Imaging: §3.5 Adaptive Gain Control Local Operator with Edge Detection

Footer

More about this content: Metadata | Version History

• How to reuse and attribute this content
• How to cite and attribute this content

This work is licensed by Tianhe Yang, Taylor Johnson, Sarah McGee, and Robert Ortman under a Creative Commons Attribution License (CC-BY 2.0).

Last edited by Taylor Johnson on Dec 25, 2006 4:25 pm US/Central.