vlfeat常见函数

---

vl_impattern

作用：
generate an image from a stock pattern.
从库存的模式中生成一幅图像。

IM=VLPATTERN(NAME) returns an instance of the specified pattern. These stock patterns are useful for testing algoirthms.
返回一个指定模式的实例。这些库存模式对测试算法非常有用。

All generated patterns are returned as an image of class DOUBLE. Both gray-scale and colour images have range in [0,1].
返回的图像的数据类型为DOUBLE，不管是灰度图像还是彩色图像，取值范围均为[0,1]。

vl_impattern() without arguments shows a gallery of the stock patterns. The following patterns are supported:
vl_impattern() 支持以下模式：

• A stock image::
  Any of ‘box’, ‘roofs1’, ‘roofs2’, ‘river1’, ‘river2’, ‘spotted’.
  这是存放在vlfeat/data中的几张测试图片。。

I = vl_impattern('roofs1') ; % 载入测试图像
figure,image(I) ; % 显示图片

• Wedge::
  The image of a wedge.
• Cone::
  The image of a cone.
• SmoothChecker::
  A checkerboard with Gaussian filtering on top. Use the
  option-value pair ‘sigma’, SIGMA to specify the standard
  deviation of the smoothing and the pair ‘step’, STEP to specfity
  the checker size in pixels.
• ThreeDotsSquare::
  A pattern with three small dots and two squares.
• UniformNoise::
  Random i.i.d. noise.
• Blobs:
  Gaussian blobs of various sizes and anisotropies.
• Blobs1:
  Gaussian blobs of various orientations and anisotropies.
• Blob:
  One Gaussian blob. Use the option-value pairs ‘sigma’,
  ‘orientation’, and ‘anisotropy’ to specify the respective
  parameters. ‘sigma’ is the scalar standard deviation of an
  isotropic blob (the image domain is the rectangle
  [-1,1]^2). ‘orientation’ is the clockwise rotation (as the Y
  axis points downards). ‘anisotropy’ (>= 1) is the ratio of the
  the largest over the smallest axis of the blob (the smallest
  axis length is set by ‘sigma’). Set ‘cut’ to TRUE to cut half
  half of the blob.

All pattern accept a SIZE parameter [WIDTH,HEIGHT]. For all but the stock images, the default size is [128,128].

---

IMAGE

用来显示图像。
image(C) displays matrix C as an image. Each element of C specifies the color of a rectilinear patch in the image. C can be a matrix of dimension MxN or MxNx3, and can contain double, uint8, or uint16 data.

---

vl_sift