Chapter 6

In chapter 5, you learned basics of editing graphical images in paint format and draw format. However, creating a freely drawn image such as handwritten cartoon figures is difficult using graphic editors. Furthermore, computers are of little help if you are not good at drawing in the first place. Scanner provides a very convenient way to convert a printed image into an electronic form so that you can use it in a graphical editor.

Scanner works similarly to a copying machine. The difference is that a scanner outputs an electronic form of the scanned image, instead of a copy of the original image.

Resolution and image type are the two most important settings you need to know when you scan a graphical image. Resolution refers to how detailed the scanner scans the image. Like printers, it is expressed in a unit called dots per inch or dpi. For example, 300 dpi means that a scanner creates an electronic grid of 300 X 300 in one square inch. This makes 90,000 individual cells in one square inch for a scanner to fetch the image. The higher the resolution setting, the more detailed the scanning will be. Figure 6.2A shows a situation where the original drawing (Figure 6.1) is scanned at a low resolution setting. Compare it to Figure 6.2B where the resolution is significantly higher. If you have a small but detailed image to scan, you should set the scanner to scan at higher resolutions. You do not have to use a copying machine to enlarge the original image because you can achieve the same effect by changing the resolution. When scanning is done at higher resolutions, the size of the file containing the result becomes larger.

Three major types of images are black & white, gray scale and color. Different terms are used to refer to these image types. Some examples are color depth, 16-bit color, color(256), RGB, etc.

(In the following discussion, the word "cell" refers to the individual cells in the grid as in the above example.)

A black & white image is composed of only two types of dots, namely, entirely black or entirely white. (See Figure 6.1.) A line drawing or line art is a good example of this type of image as shown in the above example. When a scanner is set to scan an image as black & white, each cell becomes either totally black or totally white. So if you scan a regular color photograph as a black & white image, you are likely to get an image that is entirely pitch black except for very bright areas.

A gray scale image is like a black & white photograph or black & white TV. (See Figure 6.3A.) It does not have colors but different shades of gray creates richer image. When an image is scanned as gray scale image, each cell senses the brightness of the image and chooses an appropriate level of gray. Scanners usually recognize 256 shades of gray when they are set to scan in this mode.

A color image is like a color photograph. (see Figure 6.3B.) Unlike a regular photograph which is composed of a virtually unlimited number of colors, the number of available colors needs to be set when an image is scanned by a scanner. It determines the number of color choices for each cell when a scanner goes through the image. Typical numbers you encounter among choices are 256, thousands (approximately 32,000), and millions. Thousands and millions of colors may be referred to as RGB which stands for red-green-blue. "Bit depth" is another term commonly used to refer to the number of colors. 8-bit refers to 256, 16-bit for 32,000, and 24-bit or 32-bit for millions of colors. (2-bit refers to 2 colors, namely, black & white image.) As the number of colors becomes larger, the image becomes more photo realistic. However, the file size grows large, so you must be careful about storage space on your hard disk. For our purpose, 16-bit color or 32,000 colors will be sufficient for most images

Once you understand the concepts of resolution and image types, using a scanner is simple. Find a computer with a scanner attached. You may find a separate software for scanning, or a graphic editor (e.g. Adobe PhotoShop) with a built-in scanning software. You may want to seek assistance for the first time around to get used to the operation of a particular model of scanner and its software. Figure 6.4 shows an example of a scanning software. Although different switches are available, the two most important settings, namely resolution and image type, will get you started. You should be able to figure out what other settings will do to the image as you use the scanner more.

The general procedure of scanning an image with a flat-bed scanner is as follows.

The image type is an important setting. (Figure 6.5) This is the result of the same photo shown in Figure 6.3B scanned as a line drawing.

Text scanning is done to convert printed text into electronic format so that it can be read by, say, a word processing software. Text scanning saves time for typing when the original is only available in printed format. Typed manuscripts, newspaper and magazine articles are good candidates for scanning. Scanning is obviously unnecessary if you are planning to use authentic text in its original form such as newspaper clippings. Text scanning, however, is useful when you need to modify or format the original text in order to adopt it in your instruction.

Text scanning requires two steps. First, the page with letters is scanned as it were a graphical image. Then, a software attempts to recognize each letter on the page. The entire process is called "Optical character recognition" or OCR. (Omni Page is a well known software in this category.) Most OCR software packages perform both of these two steps, but you can certainly scan a page on one machine, save it on the disk, and go to another computer to perform the character recognition. Many OCR software can handle variations in alphabetical languages. (Cyrillic and Greek letters may require special add-on packages.) Since recognizing thousands of Chinese characters is a much more demanding task than alphabetical letters and its variations, text scanning in Chinese and Japanese require specialized software packages. Figure 6.6 shows a screen of a Japanese OCR program.

The scanning process is essentially the same for text scanning and for graphical scanning. A scanner treats everything as a graphical image. Letters on a page are considered to be a collection of line drawings, so the scanner should be set to scan black & white image (2-bit). The resolution setting will vary depending on the size of letters on the page. If letters are small, the resolution setting should be higher. There is no need to enlarge the original with a copying machine since you can achieve exactly the same effect by changing the resolution.

Once a page is scanned, OCR software first tries to identify text regions and look for text lines so that it can determine the size of each letter. Some software attempts to identify text regions automatically while others ask the user to identify them. When a page contains pictures, photographs, or letters in various sizes such as headlines, software may have trouble recognizing them. You can solve the problem by running the software in two or more paths in these cases. If the scanned page is not horizontal, the software may compensate it automatically or you may need to re-scan the page. Next, an image of each letter is compared against the model image that the software has. If the image matches, the corresponding letter is output. This process results in the text which can be read into a word processing software.

The success of text scanning largely depends on how clean the original is. Recognition of handwritten document is not possible yet.

Digital cameras are gaining popularity because it is an easy way to add photos to Web pages. A digital camera captures a view through its lens, and saves it as an electronic file just like a graphical file. A camera has a resolution like a scanner. However, unlike that of a scanner, the resolution of a digital camera is fixed. You can change resolution by moving closer/away or by using a zoom function. To illustrate this point, suppose you are taking a picture of an apple. If you take a close-up shot, you are capturing the image with higher resolution. On the other hand, if you take a distant shot, you are capturing an image of the apple with lower resolution. Resolution of digital cameras vary depending on their prices. The resolutions are usually expressed either in a grid form (e.g. 640X480 or 1028X768) or in a total number of pixels (number of cells in a grid.) Common numbers we see in recent advertisements are like 1.3 millions (equivalent to 1600X1200). Since digital cameras save images in electronic form, you can easily transport them into computers.

Digital video camcorders can capture moving images in digital form. You can think of it as a series of rapid exposures of a digital camera.