Tuesday, March 20, 2007

Image and Video Representation

Information can be represented through picture, sound, or combination of both of them. When a picture of an object is taken, the object will be spatially represented through an image. Later on, this image will be represented back to the viewer through number of mechanism, e.g. digitization, encoding, formatting, etc. The same things happen with video. However, there are two kinds of video, analog video and digital video, where the latter offers many advantages over the first. Let's discuss more about image and video.


A. Image Digitizing

There are some things happen to images before human eyes can see them on screen or monitor. In the case of sound representation, samples are taken in order to represent the sound to the listener. In addition, according to Nyguist’s law, the more sample taken, the better the result will be. Thus, the same law applies for image sampling.

Firstly, imagine that there is grid on the image. Each cell of the grid represents a point where a sample is taken. The cell will represent the pixel, while the size of the grid will represent the resolution of the image. Moreover, the higher the resolution (i.e. the area of the sample/ the size of the grid), the better the image quality will be. However, a high resolution will result in huge number of samples, which consequently result in a large image file. Thus, a compression technique is needed to reduce the size of an image file.

Secondly, the result of the sample will be stored as matrix of numeric values. These numeric values represent the intensity of each sample/ pixel which will be used to convert them to monochrome, grey-scale, or color. These color depth (i.e. monochrome, grey-scale, color) will also affect the size of image file as each color depth requires different number of bits.

Finally, the size of an image file can be calculated by multiplying the image resolution by its color depth. For instance, an image with 800x600 resolutions in colors (24 bits) requires 800x600x24=11,520,000 bits or approximately 1,44 M.

B. Computer-Generated Graphics

Unlike digitized image which specified through pixel matrix, graphic is specified through graphics primitives (e.g. line, square, and ellipse) and their attributes. However, graphics also need to be converted into pixel matrix before it can be rendered on the screen. In addition, graphics requires less space than digitized image and it can be labeled.

C. Image Analysis

Image analysis is one main area of image processing. The objective of image analysis is to analyze a scene through analyzing an image in order to retrieve certain information about objects in the scene. Some application areas of image analysis:

§ Object identification and tracking

§ Image enhancement

§ Pattern detection and recognition

§ Scene analysis

§ Computer vision

D. Image Recognition

Image recognition is performed in order to locate and recognized objects and their spatial relationship in a scene by analyzing each pixel of an image. It involves several steps:

1. Formatting: Digitizing the image.

2. Conditioning: Removing noise and highlighting interesting parts of the image.

3. Labeling: Identifying and labeling structures/ primitive object (e.g. edge) or shape primitives (e.g. line).


Grouping: Grouping objects based on their labels and recording spatial event to which pixels belong.

5. Extracting: Generating a list of properties for each set of pixels in a spatial event.

6. Matching: Recognizing objects in the image by comparing each object in the image with previously stored models and determining the best matched one.


A. Digital Video

Digital video offers some advantages over analog video:

§ Interactivity

Digital video allows viewer to access any part of the video since it is stored on random access devices. For instance, when a video is viewed using Windows Multimedia Player, viewer can go directly to specific scene/ part of the movie just by clicking the “seek” timeline bar.

§ Editing

Video editor can easily edit and re-edit a digital video using inexpensive tool while still maintaining the quality of the video and without risking the storage medium.

§ Quality

Digital signals are not really affected by atmospheric condition compared to analog signals. Also, digital video has lower bandwidth than analog video. Thus, a digital video with better picture quality can be transmitted in the same bandwidth as analog video with standard picture quality.

§ Transmission/ Distribution Options

Digital video can be distributed on a cheap storage medium, such as CD or DVD. It can also be viewed directly by streaming the video through the internet or being sent as a whole as a downloadable video content/ file.

B. Analog Video

Fig. 1 describes the process of how a TV camera records a video.

fig. 1. Video capturing process

1. The TV camera catches light.

2. The incoming light is converted into electrical signals by using prism to split the incoming light into red, green, and blue (RGB) channels.

3. Each pixel point on the CCD will electrically respond to the amount of light focused on its surface and then convert them into electrical impulses.

4. The CCD chips “read” the impulse line-by-line.

On the receiver side (i.e. the TV), the impulse need to be converted back into light. In order to do so, an electron gun is used to glow phosphor lines which covers the back of TV screen.

Furthermore, there are three main elements of a standard video camera outputs:

§ Luminance

It contains information about brightness of a pixel point. It is needed for black and white TV, but it is not needed for color TV.

§ Chrominance

It contains the color information of a pixel point. This information needed by color television. It consists of two parts, hue and saturation. Hue describes the actual color displayed, while saturation describes the intensity of the color.

§ Sync

It is a series of electrical pulses that control the timing of each frame of video information.

Considering luminance and chrominance, there are two kinds of video:

§ Component Video

It is where luminance and chrominance signals are kept separate, resulting in better video quality.

§ Composite Video

A video where luminance and chrominance signals are combined (i.e. luminance interlaces chrominance), enabling black and white TV to reconstruct scenes captured by color TV.

C. Digitizing Analog Video

Analog video needs to be digitized at the same rate it is played. The video playback will appear jittery if there are significant numbers of dropped frames. Moreover, in order to digitize an analog video (i.e. capturing video and save it on disks), the input device should have Firewire or a video capture card.


1. How image resolution and color depth affect the size of an image file?

2. In context with Image Recognition, how objects in an image can be recognized?

3. What are the advantages of digital video over analog video?


[1] FC. Balok, MS_Video.pdf.

[2] FC. Balok, MS_image.pdf.

No comments: