Image Processing

lemonadeviraginityΤεχνίτη Νοημοσύνη και Ρομποτική

6 Νοε 2013 (πριν από 3 χρόνια και 7 μήνες)

78 εμφανίσεις

Image Processing
Image processing can improve the way an image looks. Image-processed data, or a
processed image helps you distinguish and recognize more subtle characteristics of the
original image.
Menu Options
File Menu
Select Quit to exit the Image Processing Demo and return to the IDL Demo main
screen.
About Menu
Select About image processing for information about the Image Processing Demo.
Features
IDL offers several ways to process an image, some of which are introduced below:
Zooming
Smaller sub-sections of large datasets can be manipulated.
The image shown here is an aerial view of New York City. A section of lower Manhattan
is enlarged using the REBIN function on a subscripted section of the original image array.
To reposition the zooming area, click your mouse on the area you would like to see more
closely.
Fourier filtering
IDLs Fast Fourier Transform (FFT) function can process vectors and 2D array in either
forward or reverse.
Filter width slider
Selects the filter width. A higher filter width gives the filtered image higher resolution.
Pixel scaling
The BYTSCL (byte scale) command scales pixel values into the range of available
colors.
Minimum slider
You can specify the minimum byte value allowed in the reconstructed image.
Maximum slider
You can specify the maximum byte value allowed in the reconstructed image.
Histogram
Histogram equalization can be used to change the visible contrast of an image.
Minimum slider
You can specify the minimum byte value allowed in the reconstructed image.
Maximum slider
You can specify the maximum byte value allowed in the reconstructed image.
Edges
The SOBEL function performs edge enhancement on an images.
Smooth width slider
Sets the width of the smoothing function. A higher width produces a reconstructed image
with lower resolution.
Dilate and Erode
IDLs Dilate and Erode functions operate on shapes within an image. In this example,
before we start dilating and eroding, a threshold mask is derived from the grayscale
intensities in the image. All values less than a given percent of the image s maximum
intensity are masked.
Break Mask
Click this button to Erode and then Dilate the shape of the mask. This has the effect of
 breaking off details in the masks shape. This operation is known as Binary
Morphological Opening.
Fuse Mask
Click this button to Dilate and then Erode the shape of the mask. This has the effect of
 fusing together parts of the masks shape. This operation is known as Binary
Morphological Closing.
Neighborhood Mins.
Click this button to perform grayscale erode and then grayscale dilate on the image itself,
where masked. This yields neighborhood minimums in the masked regions of the image.
Thus bridges and other details are removed. This operation is known as Grayscale
Morphological Opening.
Convolution
Convolutions may be performed on arrays of 1, 2, and 3 dimensions. This screen shows
the effects of convolving an image with a kernel.
Use your left mouse button to click on the squares of the kernel grid, toggling them on or
off.
Convolve button
Computes and displays the convolved image.