Computer and Robot Vision I Chapter 3 Binary
Computer and Robot Vision I Chapter 3 Binary Machine Vision: Region Analysis Presented by: 傅楸善 & 李祐賢 0983868080 r 06922085@ntu. edu. tw 指導教授: 傅楸善 博士 Digital Camera and Computer Vision Laboratory Department of Computer Science and Information Engineering National Taiwan University, Taipei, Taiwan, R. O. C.
3. 1 What’s Region l Binary: only 0 and 1. DC & CV Lab. CSIE NTU
3. 1 What’s Region l Gray level value: range from 0~255 DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l area: l centroid: DC & CV Lab. CSIE NTU
3. 2 Region Properties l Region intensity histogram: DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l Average gray level (intensity) l Gray level (intensity) variance l right hand equation lets us compute variance with only one pass DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l Eg: center is in but not in l DC & CV Lab. CSIE NTU for
3. 2 Region Properties (cont’) l Perimeter : pixels neighbors , successive P/A can be used as a measure of shape’s compactness and circularity DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l mean distance R from the centroid to the shape boundary l standard deviation R of distances from centroid to boundary DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l Haralick shows that has properties: 1. digital shape circular, increases monotonically 2. similar for similar digital/continuous shapes 3. orientation (rotation) and area (scale) independent DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l Micro-texture properties: used to measure the texture of regions. DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l texture second moment (Haralick, Shanmugam, and Dinstein, 1973) l texture entropy l texture correlation DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) where l texture contrast DC & CV Lab. CSIE NTU
3. 2 Region Properties (cont’) l texture homogeneity where k is some small constant DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (極點) DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l dierent extremal points may be coincident DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l association of the name of the eight extremal points with their coordinates DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l directly define the coordinates of the extremal points: DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l association of the name of an external coordinate with its definition DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l extremal points occur in opposite pairs: topmost left bottommost right, topmost right bottommost left, rightmost top leftmost bottom, rightmost bottom leftmost top l each opposite extremal point pair: defines an axis l axis properties: length, orientation DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l l orientation convention for the axes paired: with and with DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l distance calculation: add a small increment to the Euclidean distance DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l length going from left edge of left pixel to right edge of right pixel DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l orientation taken counterclockwise w. r. t. column (horizontal) axis DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l distance between ith and jth extremal point l average value of - 1. 12 = 1. 12, largest error 0. 294 = DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l How can we use extremal points? 1. Line’s length/orientation 2. Triangle’s base/height 3. Rectangle’s orientation DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l calculation of the axis length and orientation of a linelike shape DC & CV Lab. CSIE NTU
l calculations for length of sides base and altitude for a triangle DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l geometry of the tilted rectangle DC & CV Lab. CSIE NTU
l calculation for the orientation of an example rectangle DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) l axes and their mates* that arise from octagonal-shaped regions DC & CV Lab. CSIE NTU
3. 2. 1 Extremal Points (cont’) DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 2. 3 Example 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 connected components labeling of the image in Fig 2. 2 01 02 03 04 05 06 07 08 09 10 11 12 13 l DC & CV Lab. CSIE NTU
3. 2. 3 Example l all the properties measured from each of the regions DC & CV Lab. CSIE NTU
3. 2. 3 Example DC & CV Lab. CSIE NTU
3. 2. 2 Spatial Moments (binary) l Second-order row moment l Second-order mixed moment l Second-order column moment DC & CV Lab. CSIE NTU
3. 2. 3 Mixed Spatial Gray Level Moments (gray level) l l region properties: position, extent, shape, gray level properties Second-order mixed gray level spatial moments DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 3 Signature Properties (投影) l What’s Signature? DC & CV Lab. CSIE NTU
3. 3 Signature Properties (投影) For examples: l l l Optical character recognition. Hand gesture recognition. Used to measure other properties. DC & CV Lab. CSIE NTU
3. 3 Signature Properties (投影) l Optical character recognition DC & CV Lab. CSIE NTU
3. 3 Signature Properties (投影) l Hand gesture recognition: DC & CV Lab. CSIE NTU
3. 3 Signature Properties (cont’) l Area: DC & CV Lab. CSIE NTU
3. 3 Signature Properties (cont’) l rmin: top row of bounding rectangle l rmax; bottom row of bounding rectangle l cmin: leftmost column of bounding rectangle l cmax: rightmost column of bounding rectangle DC & CV Lab. CSIE NTU
3. 3 Signature Properties (cont’) l row centroid: l column centroid: l diagonal centroid: l another diagonal centroid: DC & CV Lab. CSIE NTU
3. 3 Signature Properties (cont’) l second row moment from horizontal projection l second column moment from vertical projection l second diagonal moment DC & CV Lab. CSIE NTU
3. 3 Signature Properties (cont’) l second diagonal moment related to l second mixed moment can be obtained from projection l second diagonal moment related to DC & CV Lab. CSIE NTU
3. 3 Signature Properties (cont’) l second mixed moment can be obtained from projection l mixed moment obtained directly from DC & CV Lab. CSIE NTU and
DC & CV Lab. CSIE NTU
3. 3. 1 Signature Analysis to Determine the Center and Orientation of a Rectangle l l signature analysis: important because of easy, fast implementation surface mount device (SMD) placement: position and orientation of parts DC & CV Lab. CSIE NTU
3. 3. 1 Signature Analysis to Determine the Center and Orientation of a Rectangle (cont’) l determine center of rectangle by corner location side lengths w, h orientation angle DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
l geometry for determining the translation of the center of a rectangle DC & CV Lab. CSIE NTU
l l partition rectangle into six regions formed by two vertical lines a known distance g apart and one horizontal line DC & CV Lab. CSIE NTU
3. 3. 1 Signature Analysis to Determine the Center and Orientation of a Rectangle (cont’) DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 3. 1 Signature Analysis to Determine the Center and Orientation of a Rectangle (cont’) l where rotation angle DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
3. 3. 2 Using Signature to Determine the Center of a Circle l l partition the circle into four quadrants formed by two orthogonal lines which meet inside the circle geometry for the circle its center and a chord DC & CV Lab. CSIE NTU
3. 3. 2 Using Signature to Determine the Center of a Circle (cont’) l circle projected onto the four quadrants of the projection index image DC & CV Lab. CSIE NTU
3. 3. 2 Using Signature to Determine the Center of a Circle (cont’) l l l each quadrant area from histogram of the masked projection positive if A + B > C + D negative otherwise where positive if B + D > A + C, negative otherwise DC & CV Lab. CSIE NTU
3. 4 Summary l region properties from connected components or signature analysis DC & CV Lab. CSIE NTU
Histogram Equalization (Homework) l pixel transformation l r, s: original, new intensity, T: transformation l T( r ) single-valued, monotonically increasing for l DC & CV Lab. CSIE NTU
DC & CV Lab. CSIE NTU
Histogram Equalization (Homework) l l histogram equalization: histogram linearization number of pixels with intensity j n: total number of pixels for every pixel if DC & CV Lab. CSIE NTU then
Histogram Equalization (Homework) l l Project due Oct. 17: Write a program to do histogram equalization DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l https: //vimeo. com/214365449 DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l 原圖(RGB) DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l RGB轉YCb. Cr DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l Range from (0, 137, 77)~(156, 177, 127) and get a binary graph. DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l Use co-occurrence to determine if it’s hand or noise. DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l Measure the centroid and get the main area. DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition l Use signature property to account which number it is. DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition DC & CV Lab. CSIE NTU
Example: Hand Gesture Recognition DC & CV Lab. CSIE NTU
End DC & CV Lab. CSIE NTU
- Slides: 83