Audio, Image and Video Processing Blog: Week 09: 23/11/2012

Lecture

Why Use Digital Image Processing?

No use of darkroom
Reduced time to produce a photo
No noxious chemicals
Perfect for experimenting time and again to achieve a certain desired effect
The variety of options available for things that can be done to an image when digitally processed are far greater than that of darkroom processing

Image manipulation
Enhancement
Transformation operations

(See figure's 1, 2, 3 and 4)

Figure 1 - Digital Imaging Systems

Figure 2 - Digital Camera Image Capture

Figure 3 - Sensor Spatial Resolution

Figure 4 - Digital Camera Colour

The light collected by the lens is passed through an optical low-pass filter before it is focused onto the sensor array. This optical low pass filter serves to:

Exclude any picture data that is beyond the sensor's resolution
Compensate for false colouration and RGB Moiré caused by drastic changes of colour contrast

As in pictures of thin stripes and fine point sources

Reduce infrared and other non-visible light which disturbs the sensor's imaging process

What Is Moiré and How Do You Prevent or Remove It? (See figure 5)

Figure 5 - Before and After - Removal of Moiré

"Moiré is a repetitive pattern of wavy lines or circles that occasionally appear on objects in digital captures. It occurs when fibers or fine parallel details in an object match the pattern of the imaging chip in the camera. Some camera companies incorporate anti aliasing filters in the cameras to slightly blur the tiny detail areas of captures, but other manufacturers prefer not to use them because they don’t want to sacrifice image sharpness. With or without the filtering, every digital camera is capable of creating moiré." - From PowerPoint lecture

(See figure's 6 and 7)

Figure 6 - Digital Image Fundamentals

Figure 7 - The Pixel

Bit-Mapped Graphics

An ordered array of groups of bits. That is the definition of a bit-mapped colour image when represented in a digital memory.

The colour of a single pixel on the screen is coded by these single groups. For example: 8 bits are needed for each of the red, green and blue levels. This means 24 bits are needed per pixel.

"The array of pixels could be 640 x 480 (i.e. VGA spatial resolution), with the colour of each pixel represented by a group of 24 bits requiring,

640 x 480 x 24 = 7372800 bits

i.e. approximately 7.4Mb per image. 1 Mb = 10⁶ bits" - From PowerPoint lecture

(See figure's 8, 9, 10, 11 and 12)

Figure 8 - Dynamic Range

Figure 9 - Example of an Image Using 1-Bit Colour Depth

Figure 10 - Example of 8-Bit Depth - Grey Scale

Figure 11 - Example of 8-Bit Colour Depth

Figure 12 - Example of 24-Bit Colour Depth

Colour Palette

If a computer system and pre-determine and establish what colour palette is to be used, then the system palette colours (for example 256) are used for all images.

By selected the closest 256 colours closest to those colours in an image is how the fidelity of an 8-bit (256 colour) image is often enhanced.

When using a system that can only display 256 colours, displaying multiple images simultaneously can become a problem when using the adaptive palette.

Whether or not the palette is appropriate for certain images, the system MUST choose one palette and apply it to all images shown. (See figure's 13, 14 and 15)

Figure 13 - Example of a Non-Optimised Palette of 256 Colours

Figure 14 - Example of Optimised Palette of 256 Colours

Figure 15 - Example of 256 Colour Palettes

What is Digital Image Processing?

"Four common categorisations of DIP operations are analysis, manipulation enhancement and transformation.

Analysis operations provide information on photometric features of an image e.g. colour count, histogram.

Manipulation operations change the content of an image e.g. flood fill, crop. Input image yields output image.

Enhancement operations attempt to improve the quality of an image in some sense e.g. heighten contrast, edge enhancement. Input image yields output image.

Transformation operations alter the image geometry e.g. rotate, skew. Input image yields output image." - From PowerPoint lecture

(See all figure's below) - Refresher note for blog readers: I would type out a lot more in my own words, but when there is a picture on the PowerPoint lecture slide, I just put the who slide in the blog because the whole slide is a picture all in its own right, not a slide with words and a picture.