Histogram Equalization

Building the Cantata Workspace

1. Read the image cathedral.viff and display it using the operators User defined and Display Image.
2. Determine the histogram of the image and plot it. Use operators Histogram and 2D Plot.
3. If the cumulative histogram is not an option from the Histogram operator, you can find the cumulative histogram by linear convolving the histogram with a 256-height constant kernel. Please check the list of Known Bugs and Suggestions.
4. Normalize the cumulative histogram to the desired range of the output image, e.g. 0, 255. Use the Normalize operator.
5. Convert the normalized cumulative histogram to a MAP segment (color table), using the operator Copy from Value.
6. Use the color table in the three possible ways:
   1. into the Display Image operator directly.
   2. combining the MAP segment to the original image file
   3. mapping the MAP segment to the original image
7. Use the operator Equalize to perform the histogram equalization directly. Check if your implementation generates the same results as this operator.

Exercises

• Apply this technique only to a selected portion of the image. Use the MASK segment to select the area of interest.
• One solution to obtain a more uniform histogram in the resultant image is by converting the image to floating point pixels and adding an uniform noise ranging from [0, 1] to each pixel. Equalizing this image which has larger number of gray levels will result in a more uniform histogram.