Image Processing Techniques

As has just been established, a number of factors can adversely affect RTR image quality. With the use of image enhancement techniques, the difference in sensitivity between film and RTR can be decreased. A number of image processing techniques, in addition to enhancement techniques, can be applied to improve the data usefulness. Techniques include convolution edge detection, mathematics, filters, trend removal, and image analysis. The various image enhancements and image processing techniques will be introduced in this section. Computer software programs are available, including some or all of the following programs:

Enhancement programs make information more visible.

  • Histogram equalization-Redistributes the intensities of the image of the entire range of possible intensities (usually 256 gray-scale levels).
  • Unsharp masking-Subtracts smoothed image from the original image to emphasize intensity changes.

Convolution programs are 3-by-3 masks operating on pixel neighborhoods.

  • Highpass filter-Emphasizes regions with rapid intensity changes.
  • Lowpass filter-Smoothes images, blurs regions with rapid changes.

Math processes programs perform a variety of functions.

  • Add images-Adds two images together, pixel-by-pixel.
  • Subtract images-Subtracts second image from first image, pixel by pixel.
  • Exponential or logarithm-Raises e to power of pixel intensity or takes log of pixel intensity. Nonlinearly accentuates or diminishes intensity variation over the image.
  • Scaler add, subtract, multiply, or divide-Applies the same constant values as specified by the user to all pixels, one at a time. Scales pixel intensities uniformly or non-uniformly
  • Dilation-Morphological operation expanding bright regions of image.
  • Erosion-Morphological operation shrinking bright regions of image.

Noise filters decrease noise by diminishing statistical deviations.

  • Adaptive smoothing filter-Sets pixel intensity to a value somewhere between original value and mean value corrected by degree of noisiness. Good for decreasing statistical, especially single-dependent noise.
  • Median filter-Sets pixel intensity equal to median intensity of pixels in neighborhood. An excellent filter for eliminating intensity spikes.
  • Sigma filter-Sets pixel intensity equal to mean of intensities in neighborhood within two of the mean. Good filter for signal-independent noise.

Trend removal programs remove intensity trends varying slowly over the image.

  • Row-column fit-Fits image intensity along a row or column by a polynomial and subtract fit from data. Chooses row or column according to direction that has the least abrupt changes.

Edge detection programs sharpen intensity-transition regions.

  • First difference-Subtracts intensities of adjacent pixels. Emphasizes noise as well as desired changes.
  • Sobel operator-3-by-3 mask weighs inner pixels twice as heavily as corner values. Calculates intensity differences.
  • Morphological edge detection-Finds the difference between dilated (expanded) and eroded (shrunken) version of image.

Image analysis programs extract information from an image.

  • Gray-scale mapping-Alters mapping of intensity of pixels in file to intensity displayed on a computer screen.
  • Slice-Plots intensity versus position for horizontal, vertical, or arbitrary direction. Lists intensity versus pixel location from any point along the slice.
  • Image extraction-Extracts a portion or all of an image and creates a new image with the selected area.
  • Images statistics-Calculates the maximum, minimum, average, standard deviation, variance, median, and mean-square intensities of the image data.