0

0

1889 days ago,
474 views

PowerPoint PPT Presentation
Picture Similarity Longin Jan Latecki CIS Dept. Sanctuary Univ., Philadelphia latecki@temple.edu

Image Similarity Image based, e.g., distinction of benefits of comparing pixels Histogram construct Based with respect to closeness of items contained in pictures, requires picture division

Mathematical Representation of Images A picture is a 2D flag (light power) and can be spoken to as a capacity f ( x, y). c oordinates ( x, y) speak to the spatial area of p oint ( x, y) that is called pixel ( pi cture e lement) estimation of f ( x, y) is the light power called dark esteem (or dim leve l) of picture f · Images are of two sorts: constant and discrete A ceaseless picture is a component of two factors, that take values in a continuum. E .g. : The power of a photographic picture recorded on a film is a 2D work f ( x, y) of two genuine esteemed factors x and y.

· A discrete picture is an element of two factors, that take values over a discrete set (a number lattice) E .g. : The power of a discretized 320 x 240 photographic picture is 2D work f ( i , j ) of two number esteemed factors i and j . In this manner, f can be spoken to as a 2D lattice I [ 320,240 ] A shading picture is normally spoken to with three grids: Red[ 320,240 ], Green[ 320,240 ], Blue[ 320,240 ]

Pixel based picture closeness Let f and g be two dark esteem picture capacities.

Let an and b wager two pictures of size w x h . Give c a chance to be some picture attributes that doles out a number to every picture pixels, e.g., c(a,x,y ) is the dim estimation of the pixel. Pixel to pixel contrasts:

We can utilize factual mean and fluctuation to add solidness to pixel to pixel picture distinction:

Let v(a) be a vector of all c(a,x,y) values relegated to all pixels in the picture a . Picture similitude can be communicated as standardized inward results of such vectors. Since it yields greatest qualities for equivalent casings, a conceivable difference measure is

Image histogram is a vector If f :[1, n]x[1, m] �� [0, 255] is a dim esteem picture, then H ( f ) : [0, 255] �� [0, n*m] is its histogram, where H ( f ) (k) is the quantity of pixels (i, j) to such an extent that F(i, j)=k Similar pictures have comparative histograms Warning: Different pictures can have comparable histograms

Image Histogram (3, 8, 5)

Hg Hr Hb

Histograms Histogram Processing 1 4 5 0 3 1 5 1 Number of Pixels dim level

Histogram-based picture likeness Let c be some picture attributes and h(a) its histogram for picture a with k histogram receptacles.

Homework 5 Implement in Matlab a basic picture internet searcher (no GUI required). Just look at the execution of no less than two picture separates on a little arrangement of pictures.

SPONSORS

No comments found.

SPONSORS

SPONSORS