Hardware for digital image processing

The input component is a signal source for image processing. The quality of this signal will affect the overall processing effect. The most important problem is that the input device introduces noise. Therefore, any image processing system does a lot of work to remove this noise. For example, under the meteorological satellite receiving antenna, a low-noise amplifier that needs to be frozen with liquid nitrogen before entering the image processing system is usually used in the system as an image input component. Its noise index is a key indicator that affects the work of the image processing system.

The low noise of high-end cameras improves the grayscale resolution of the image, so that the weak grayscale change information in the image can be obtained. For example, the grayscale resolution of the optical vision system requires 256 levels, that is, 28. It has extremely high requirements for input noise control, which is often treated as a special technical problem to be solved. The second is spatial resolution, which depends on the application requirements. As a general image processing system, the commonly used spatial resolution is 512×512. In addition, when the image is obtained, the input signal of the A/D conversion should meet the requirements of strict lock with the computer clock.

Common input devices are of the following types.

· Optical-mechanical type – there are optical drums, laser scanners, optical readers and microdensitometers, etc.

· Electron beam type – there are television camera tubes, flying spot scanners and color flying spot scanners, etc.

· Solid-state devices – input devices based on charge-coupled devices (CCD cameras) are widely used. They have the advantages of high spatial resolution, good position linearity and small size. In addition, compared with electron beam cameras, its outstanding advantage is that it can work under low illumination.

· Image storage components.

The selection of image storage components should meet the following requirements.

· The three-terminal storage body structure should be suitable for fast access.

· Large capacity – usually a 512×5l2×8 bit image is stored as a body, and the image storage requires many such bodies, so the capacity should be large.

·Address data management should be flexible – in addition to the three-terminal requirements, it can also be accessed by bit, by surface, and locally. At the same time, the AND, OR, addition, subtraction, splicing, up and down, left and right scrolling of each part of the storage body should be flexible. The flexibility of the search management of the memory indicated by CRT should also be considered.

(2) Image processor

The image processing system based on software processing has high requirements for the speed, memory, and function of the computer. If the speed does not meet the requirements, the algorithm should be changed or replaced with a higher-end computer. For the image processing system of the image computer type, the requirements for various indicators of the computer can be lower. However, both types require the computer to be able to expand a variety of peripherals, such as tapes, disks, CDs, mouse rolling balls, and digitizers, so as to make the image processing operation more flexible.

(3) Image output device

The purpose of the output device is twofold: one is to display the image before and after processing or the intermediate results of processing for analysis, recognition and interpretation; the other is to make a hard copy or record it in the form of data for permanent preservation. Most intermediate display devices use CRT display devices. When the resolution is low, the CRT display of the TV is used. When the resolution exceeds that of a general TV, a specially designed display is often used. Liquid crystal display devices have quickly occupied the market with their advantages such as small size and easy portability. The spatial resolution and grayscale resolution of CRT liquid crystal displays are still the two main selection indicators.

The output devices used as hard copies include flying spot scanners, optical drum scanners, laser scanners, and various printing and inkjet output devices. The grayscale resolution and color vividness of these devices have reached a practical level. In addition, hard copies of image processing can also be obtained by video recording, photography, and ordinary cameras with one-shot imaging.

(4) Typical image processing system structure

• The original image data enters the image processor through the image input component. If the original image input is an analog signal, the continuous analog image signal must be converted into a discrete digital image signal through an image acquisition card.

• That is to say, the image input to the computer is an analog signal or a digital signal, but the signal processed by the image acquisition card must be a digital signal, and its output signal can be directly received by the computer.

• After the digital signal input from the image acquisition card enters the microcomputer, the expansion slot of the microcomputer is plugged with an acquisition card with a frame memory. The frame memory receives the data input from the image acquisition card and then sends it to the computer’s memory. Then the computer memory starts to process the image data in the memory.

• When the data processing is completed, the computer will send the processed data back to the frame memory, and at the same time send the result to the display to facilitate the user to monitor the processing result.

For acquisition cards that directly use memory, it only needs to exchange data with the memory. The larger the computer’s memory and the faster the CPU’s computing speed, the faster the image processing speed. After the frame memory obtains the processing results from the computer, it transmits its data to the output component of the image. Its schematic diagram is shown in the figure.

Figure Typical image processing system structure

Image output is the ultimate goal of image processing. From a broad perspective, the output forms of images can be divided into two types: one is to make judgments or perform adjustments based on the results of image processing, for example, in quality inspection, the patch head of the optical vision system adjusts the spatial posture of the patch head based on image processing.

Of course, its output does not necessarily take the image as the final form, but only needs to be prompted for people or machines to choose. This prompt can be computer screen information, the level of electrical signals, and such output is often used in mature research applications; the other is to use images as output, which includes monitoring of the intermediate process and the output of the result image.

In short, in the hardware of digital image processing, the speed of image processing is directly related to the performance of the hardware, and of course, it is also inseparable from the processing speed of the image processing software.