User`s guide

R2014b
2-6
registration that work at a global level, applying the same mathematical transformation
to every pixel in an image. Now the toolbox supports a function that works at a
local level, capable of applying different transformations to every image pixel. The
imregdemons function returns the warped image and a displacement field that describes
how each pixel is transformed.
Image warping using displacement fields
The imwarp function now accepts the displacement field returned by the imregdemons
as input. The imregdemons function returns a warped image and a displacement field
that describes how each pixel is transformed. If you want more detailed control of a
nonrigid registration, you can pass this displacement field to imwarp to perform the
transformation. For example, by using imwarp, you can specify the interpolation method
used.
Image segmentation using the Fast Marching Method algorithm
The toolbox includes a new function, imsegfmm, that segments an image using the Fast
Marching Method (FMM) algorithm. To segment an image, you must first create a weight
image using either the gradientweight or graydiffweight functions. These functions
create an image in which every pixel is a value. You then pass this weight image to
imsegfmm, specifying where the algorithm should start, the seed location.
Image comparison using mean-squared error
The toolbox includes a new function, immse, that calculates the mean-squared error.
Color space conversion functions
The toolbox includes several new functions to convert between the RGB, XYZ, and L*a*b
color spaces.
rgb2lab rgb2xyz lab2xyz
lab2rgb xyz2rgb xyz2lab
Use these new conversion functions instead of using makecform with these conversion
types: lab2srgb, srgb2lab, srgb2xyz, xyz2srgb, lab2xyz, and xyz2lab.