Ch. 31 Flashcards
What are the 3 general approaches to processing any digital image?
Spatial, Intensity, and Frequency Domain
Point, area, and global processing, and the application of kernels, are all operations in what domain?
Spatial Domain
In which of the general digital processing domains is the image sorted out according to pixel values?
Intensity Domain
Which of the general digital processing domains relates to the objects in the image rather than the pixels?
Frequency Domain
The zoom or magnification feature is an example of (point, area, or global) processing?
Area Processing
Every image begins and end with ? domain.
Spatial
As pixels are sorted into other domains, the computer keeps track of their ? locations, so they can be placed black.
Matrix
Sorting an image by the intensities of its pixels results what?
A histogram
Sorting an image by object ? results in a frequency distribution.
Size
The histogram plots the number of ? against each pixel value.
Pixels
The frequency distribution plots the number of objects or details against their ?
Size
While the image is separated into the intensity domain (histogram), specific ? can be targeted to be altered.
Densities
While the image is separated into the frequency domain, specific sizes of objects or details can be targeted for ? or ?
Suppression or Enhancement
When intensity or frequency operations are complete, the changed data are placed back into the pixels of the ? matrix to reconstruct the image.
Spatial
For gradation processing, the rescaled data set is fed into an anatomical ? that was determined when the operator selected the procedure at the console.
LUT
When a gradient curve is plotted on a graph showing the various densities in an image, the average brightness of the image is represented by the left to right ? of this curve, and the contrast as how steep the curve is.
Position
A look up table really is a simple table with two columns: One for ? and one for ?
Input or Output
The conversion of input values to out values is generated by mathematical ? that can be represented as a ? curve on a graph.
Formula; Function
The types of formulas used for gradation processing are referred to as ? transformations.
Intensity
If the dynamic range or bit depth of a digital processing system is too limited, it is possible for data ? to occur when either brightness or contrast adjustments are made.
Clipping
Bit depth is the range of different gray levels for a computer, LCD monitor or other ? devices.
Hardware
? Is the range of different gray levels made available by a computer system including it’s installed software.
Dynamic range
Gray scale is the range of different gray (brightness) levels present in a ? image.
Displayed
With conventional radiography,when soft tissue techniques were used to demonstrate soft tissue areas, the ? were then depicted too light.
Bones
In dynamic range compression, compressing the gray scale curve brings extremely light densities in the image up to a ? level, and at the same time extremely dark areas are made ?
Darker; Lighter
Since the actual gray scale of the image does not use up to the entire dynamic range in the system, the ? image can still be darkened or lightened as a whole
Displayed
Since the bit depth of a computer can far exceed the range of human vision memory ? can still be darkened or lightened as a whole.
Storage space
Mathematically, DRC finds the mid-point of the gray scale curve, (the average brightness or density level), then progressively ? pixel values above this point, and progressively ? pixel values below it.m
Reduces; Increases
Applied to a degree visibly affecting the image, DRC results in tissue ? which takes out the darkest and lightest densities in the image.
Equalization
Tissue equalization or contrast equalization can simulate the traditional ? technique.
Soft tissue
Tissue equalization is actually gray scale truncation - Elimination of darkest and lightest densities which results in a ? looking image.
Grayer
Detail processing is characterized by its ability to treat fine details as a ? component of the image, without changing the overall brightness or contrast.
Separate
Detail processing can be performed either in the ? domain or in the ? domain.
Frequency; Spatial
For the first step in unsharp mask filtering, a mask image is created which contains only the ? structures in the image.
Gross
The “unsharp” mask is not really geometrically blurred; it appears blurry or unsharp because the ? details have been removed through averaging.
Finer
Structures that are smaller than the kernel size are ? and no longer visible in this mask image.
Suppressed
The larger the size of the ? matrix used, the wider the region used for averaging, and the more “blurred” the mask image appears.
Kernel
In the second step for unsharp mask filtering a ? of the “unsharp” mask is created by image reversal.
Positive
In step 3 for unsharp mask filtering, the positive mask is effectively superimposed over the original image, such that positive and negative pixel values cancel each other out for anything present on ? images.
Both
The net result is ? (High-pass filtering)
Edge enhancement
Edge enhancement features such as blurred mask subtraction should not be over-used, because they also enhance the levels of ? in the image.
Noise
The application of too small a kernel may remove ? from the image such that diagnostic information is lost.
Details
Just the opposite of unsharp mask filtering, kernels can also be used to suppress image ?
Noise
? is a form of low-pass filtering that softens edges and reduces noise.
Smoothing
Excessive noise reduction can lead to ? of details.
Loss
In an image that already has ? contrast, applying smoothing can lead to loss of detail.
Low
Kernels work better for suppressing ? noise, such as quantum mottle, whereas frequency processing works better for periodic noise.
Random
An image can be represented not only as a collection of pixels in space, but also as a collection of waves with different ?
Frequencies
3 steps for final preparation of image display:
- Noise reduction
- Additional gradation processing
- Formatting for display
Because edge enhancement, used universally for detail processing, results in increased noise in the image, noise reduction must ? be applied at this stage.
Again
Two general types of noise are ? noise and ? noise.
Random; Periodic
Periodic noise consists of artifacts that tend to be of roughly ? size and occur in a regular ?. An example is electronic “snow”
Consistent; Pattern
By reducing noise in any form, all-important ? is enhanced.
Signal - to - noise ratio
Frequency processing is ideal for removing electronic an other periodic forms of noise, because these artifacts all occur in the ? image detail layer, while normal anatomy is spread across several image layers.
Same
Kernels work better for suppressing ? noise, such as quantum mottle, that occur in several image detail layers.
Random
? Is defined as the difference in signal contribution (intensity) between two different image densities or specific tissue areas, divided by the background noise level.
Contrast - to - noise ratio (CNR)
Formula for CNR:
CNR = T1-T2/N
Noise levels can be measured with test objects presenting ? between lines or spots.
Just Noticeable Differences
Some manufacturers “calculate” a “CNR image” for subtraction from the overall image to reduce noise. In reality, the size of the ? or noise artifacts is determined, and the frequency layer is deleted.
Granularity
