Lec 3
Lec 3
● Image acquisition
Last Lecture ○
○
Single sensor
In-line sensor
○ Sensor array
● Sampling and quantization.
Digital Image Acquisition Process
2D Imaging Sensor
Image Acquisition
FIGURE: CCD
Sampling
Bilinear Interpolation:
Quantization Effect
Quantization Effect
f(x,y) = i(x,y)r(x,y)
□i(x,y):
determined by the nature of the light source
bounded by
0 < i(x,y) < ∞
□r(x,y) :
determined by the nature of the objects
bounded by
0 < r(x,y) < 1
Gray Level
3
Digital image representation
● Visual Intensity array.
● more common.
● Intensity of each point is proportional to
the value of f at that point.
● Eg-only three equally spaced intensity
values.
● Normalized to [0,1],values can be either
0, 0.5, 1.
● A monitor /printer converts these values
to either black, gray or white
respectively.
4
Digital image representation
● Displaying the numerical values of
as a array.
● For large images ,complete array values
to be displayed is tedious and nothing
much can be inferred from it.
● Only parts of the image are printed as
numerical values.
5
● Continuous image to Digital image. ● Terms - Dynamic range.
● ● Ratio of maximum measurable intensity
● Matrix representation- M rows & N to the min detectable intensity level.
columns. ● Saturation and noise.
● For Image digitization –values of M,N ● Highest value beyond which all intensity
and L (the number of discrete intensity values are clipped.
levels) need to be chosen.
● M,N –positive integers.
● L value- depends on digital storage and
quantizing hardware considerations
● L is taken as integer power of 2
6
7
Image Contrast
● Difference in intensity between the
highest and lowest intensity level in an
image.
● Contrast ratio – highest/lowest intensity
level in an image.
● Image with high dynamic range→ expect
high contrast.
● Image with low dynamic range→ dull
washed out gray look.
K – bit image
8
Spatial resolution
● Spatial resolution :
○ measure of smallest observable detail in an image.
○ Line pairs per unit distance
○ Pixel (dots) per unit distance.
● dpi (dots per inch) unit in printing and publishing industry.
○ Newspaper -75dpi
○ Magazines - 133 dpi
○ Glossy brochure -175dpi
● Higher Image size means better image ??
○ 1024x1024 image vs 512x512
9
● Image size alone doesn’t mean ,it is a better image. The spatial resolution has to be considered also.
10
Understanding resolution
● Pixels
● The amount of these pixels and the way
they are distributed are the two factors
that you need to consider to understand
resolution.
● Pixel count
● Pixel density
● “a rubber band, you can stretch it or
shrink it but you’re not changing the
composition of the band, you’re not
adding or cutting any of the rubber.”
Effect of reducing spatial
resolution
● 1250,300,150 and 72 dpi.
● Slight distortion in the large black
needle in the first two images.
● As the resolution reduces the
difference in quality or degradation of
the image is more visible.
15
Intensity resolution
● Intensity resolution : Smallest observable change in intensity level.
● Number of bits used to quantize intensity.
● The number of intensity levels is an integer power of 2 .
● Image that is quantized into 256 levels has 8 bit intensity resolution.
● Most common no of bits is 8 bit.
● Capturing small levels of brightness : intensity resolution high.
● For a Intensity profile
● Coarse discretization ,smooth variation /transitions will not be detected properly.
16
Effect of intensity resolution reduction
18
Ǫuantity equals quality ??
● People think that megapixels equal quality
● On top of the quantity you should also consider the depth of the pixels, this is what determines
the amount of tonal values that your image will have.
● For example,
○ a 2-bit depth can store only black, white and two shades of grey,
○ With an 8-bit photo (2 to the power of 8 = 256) you’ll have 256 tones.
Role of image size