Redundancy
Redundancy in images refers to the presence of unnecessary or repetitive
information that can be removed without significantly affecting the image
quality. This is essential in image compression techniques. The main types
of redundancies in images are:
1. Spatial Redundancy (Inter-Pixel Redundancy)
a. Definition: This occurs when neighboring pixels in an image
have similar intensity values, leading to redundant
information.
b. Reason: Most natural images contain large smooth regions
with gradual intensity changes, except at edges.
c. Example: In a grayscale image of a sky, adjacent pixels may
have very similar intensity values. Storing all these similar
values increases file size unnecessarily.
d. Compression Techniques:
i. Run-Length Encoding (RLE): Groups consecutive
identical pixel values into a single value with a count
(e.g., "AAAAABBBB" → "5A4B").
ii. Differential Pulse Code Modulation (DPCM): Stores
the difference between consecutive pixels instead of
absolute values.
e. Usage: Lossless image compression techniques like PNG and
GIF exploit spatial redundancy.
2. Spectral Redundancy (Inter-Channel Redundancy)
a. Definition: Exists in color images due to similarities between
different color channels (e.g., Red, Green, Blue in RGB).
b. Reason: Some information in one color channel can often be
predicted from another (e.g., Green and Blue values are often
similar in nature scenes).
c. Example: In an RGB image, if a pixel has high red intensity, it
might have a correlated green intensity.
d. Compression Techniques:
i. Color Space Transformation: Convert RGB to YCbCr,
where Y (luminance) contains most of the important
information, reducing redundancy in color channels.
ii. Principal Component Analysis (PCA): Removes
redundant spectral information by transforming color
data into uncorrelated components.
e. Usage: JPEG compression reduces spectral redundancy by
applying transformations like YCbCr.
3. Temporal Redundancy (Inter-Frame Redundancy in Videos)
a. Definition: In video sequences, consecutive frames often
contain the same objects with minor movement.
b. Reason: Since most video frames do not change significantly
from one frame to another, storing every frame separately
wastes storage.
c. Example: In a surveillance video, a stationary background
remains the same in multiple frames, making full storage
inefficient.
d. Compression Techniques:
i. Inter-frame Compression: Stores only the changes
between frames rather than the full frame (e.g., MPEG,
H.264).
ii. Motion Compensation: Detects moving objects
between frames and encodes only the differences
instead of full frames.
e. Usage: Used in video compression standards like MPEG,
H.264, and HEVC (H.265).
4. Psycho-Visual Redundancy
a. Definition: Based on the fact that the human eye is more
sensitive to brightness details than color details.
b. Reason: Human perception is less sensitive to high-frequency
details and small color variations.
c. Example: Removing high-frequency details in an image will
not be noticeable to the human eye but reduces file size.
d. Compression Techniques:
i. JPEG Compression: Uses Discrete Cosine Transform
(DCT) to remove less noticeable details in images.
ii. Chroma Subsampling (4:2:0, 4:2:2): Reduces the
resolution of color channels while keeping luminance
intact.
e. Usage: Applied in lossy compression techniques like JPEG and
MPEG video formats.
5. Coding Redundancy
a. Definition: Occurs when inefficient encoding assigns longer
codes to frequently occurring pixel values.
b. Reason: Some pixel values occur more frequently than
others, yet all are stored with the same number of bits.
c. Example: In a grayscale image, if most pixels have intensity
values around 128, storing them with 8-bit encoding wastes
space.
d. Compression Techniques:
i. Huffman Coding: Assigns shorter codes to frequently
occurring pixel values and longer codes to less frequent
values.
ii. Arithmetic Coding: Replaces fixed-length codes with
variable-length codes based on probability distribution.
e. Usage: Used in lossless compression formats like PNG, GIF,
and JPEG 2000.
