computer vision 机器视觉 知识点整理

名词解释

hyper-column（出现两次）

a region of primary visual cortex (V1) that contains neurons covering the full range of RF types for a single spatial location

inpainting

the filling-in of missing or corrupted parts of an image

detector(出现两次)

the detector is the method used to locate image features or interest points which are suitable for matching.

descriptor(出现两次)

descriptor is an array of feature values associated with each interest point. There descriptors are compared to determine which points match.

quantization

is the process of representing light intensity using a finite number of discrete values

pixelization

is the process of representing light intensity at a finite set discrete locations in a grid.

segmentation

segmentation is the process of partitioning the pixels in an image into different groups that correspond to different objects or features.

region-based segmentation Algorithms

K-means ; region growing ; region merging ; split and merge

image processing（出现两次）

signal processing applied to an image, with another image as the resulting output

mid-level vision

a range of processes that group together related image elements, and to segment them from all other image elements

horopter

an imaginary surface on which all points have zero disparity

aperture problem

the aperture problem refers to the fact that the direction of motion of a small image patch can be ambiguous

geons

geons are geometrical icons, simple volumes such as cubes, spheres, cylinders, and wedges.
There is a hypothesis that object recognition in biological systems is based on the ability to recognize a small set of shapes geons from which more complex objects are built up. The visual system breaks down an object into geons and compares this arrangement of geons with arrangement of geons of known objects.

computer vision

extracting information about the world from images

down-sampling

reducing the sampling rate, or resolution, of an image

transduction

the transformation of one form of energy to another

LGN Lateral Geniculate Nucleus

Lateral Geniculate Nucleus is a point along the visual pathway between the retina and the primary visual cortex

centre-surround receptive field

A receptive field is an area of visual space
A centre-surround receptive field is one that detects a difference between the centre of the area and the outside of the area. Thus a neuron with a centre-surround receptive field will respond wither if the centre is brighter than the surround or if the surround is brighter than the centre.

correspondence problem

The correspondence problem is the problem of finding the same 3D location in two or more images.

group and segment （the role of mid-level vision）

epipolar constraint

The epipolar constraint can be used to reduce the search space for possible matches when solving the correspondence problem: it uses geometrical information to reduce the search space to a line.

其他概念

必备公式

3Dworld中的物体到相机中的点的公式

pinhole camera model

depths of the image and obejct

coplanar stereo计算深度的公式

也可以用在video sequence中
当camera 沿着z轴移动的时候：

这个也可以从原来的公式中推导出来，就是f被x1*（pixel->mm）替换了
time-to-collision的计算公式：

R of Harris detector

metrics

cross-correlation

normalised cross-correlation

correlation coefficient

sum of absolute difference

其他

计算卷积 1

colour opent cells

Gestalt laws

• proximity
• similarity
• closure 会消失一部分，让人感觉到
• continuity ： straight and smoothly curving lines
• common fate:
• symmetry
• common region
• connectivity
  

global and local

camera

sampling the Image In a camera and In an eye (出现两次)

focusing a camera and an eye

advantage of LOG （出现四次）

hough transform

---

待补充 K-mean cluster

---

k-nearest neighbour classifier

这个与K-mean cluster不相关

hierarchical clustering algorithm

region growing algorithm pseudo-code(出现两次)

region merging algorithm pseudo code

split and merge algorithm pseudo-code

splitting and merging works well if the regions are farily homogeneous. If not, many spurious regions are created.

Normalised Cuts pseudo-code

tamplate matching pseudo code

computer vision Is an “ill-posed” problem

Ill-posed, forward problem

Ill-posed Inverse problem

CCD camera forms an RGB Image

stereo correspondence problem

Ill-posed problem and constraints (出现两次)

correlation-based and feature-based methods (出现两次)

coplanar Image calculation

use of depth Information

识别系统需要insensitive的(出现两次)

• viewpoint
• lighting(illumination)
• non-rigid deformations
• occlusion
• within category variations

sliding windows

viewer-centred and object-centred

calculate a Gaussian by hand

Canny edge detection algorithm pseudo-code（出现两次）

RANSAN pseudo code

aperture problem and how to overcome it

constraint of video correspondence problem

cues to depth

Gaussian\Laplacian pyramid

demosaic

V1（出现两次）

how V1 work with ganglion cells

Lateral Geniculate Nucleus

role of Harris corner detector

pseudo-code of bag-of-words

correspondence problem and stereo vision

MATLAB代码记忆

未解决的问题