CH2. Point Processes Arithmetic Operation Histogram Equalization <Experiment 4> Contrast Stretching Intensity Transformation <Experiment 5>
Arithmetic Operation What happens if we add, subtract, multiply, divide the pixel value by a constant. Let two images be X[i,j] & Y[i,j], and obtain aX[i,j]+(1-a)Y[i,j], 0 < a < 1. LUT(Look-Up Table) Operation: fastest pixel operation
Histogram Equalization Image histogram: distribution of pixel values, probability density function (Photoshop: info. Histog.) Histogram Equalization: Modify the histogram to be uniform -> image enhancement for dark or bright area (Photoshop: image -> adjustment -> equalization) Histogram Equalization for 0-255 grey image Count # of 0-255 grey pixels & save in {int h[0]-h[255]} Obtain the CDF of h[0]-h[255] & save in {int cdf[0]-cdf[255]} cdf[i]=cdf[i]*255/cdf[255] : normalization image x[i][j] = cdf[(int)x[i][j]] : LUT method If [ slope of cdf = 1] no changed pixels, If [ slope of cdf < 1] larger range of pixels occupy smaller range If [ slope of cdf > 1] smaller range of pixels occupy larger range
<Experiment 4> Open Lena.raw & Baboon.raw images, & obtain the images of aX[i,j]+(1-a)Y[i,j], for a=0.25, 0.5, 0.75. Develop your own program for the histogram equalization. Compare your results with the Photoshop’s
Contrast Stretching Rule: high, low: The maximum & the minimum pixel value in an image
Histogram Equalization vs. Contrast Stretching Observe the histograms of Lena_dark.raw, Lena_bright.raw, Lena_dark_bright.raw. Compare the performances of Histogram Equalization & Contrast Stretching
Intensity Transformation Implement by using LUT Operation: photoshop, image>adjustments>curves Gamma Correction & 0-255 normalization Compensate the nonlinearity between cameras & monitors Ex) RGB monitor: gamma = 1.4 – 2.8
Intensity Transformation Gamma Correction
Intensity Transformation Posterizing: quantization photoshop, image>adjustments>posterize
Intensity Transformation Thresholding: binalization, two-tone imaging photoshop, image>adjustments>threshold
Intensity Transformation Bit clipping & contrast stretching
Intensity Transformation Iso-intensity contouring & range highlighting:
Intensity Transformation Solarizing: photoshop, filter>stylize>solarize
Intensity Transformation Parabola transformation:
<Experiment 5> Develop your own program for Contrast Stretching & compare the results with those of Histogram Equalization Implement all the LUT operations introduced in this lecture note.