• Image Processing
• Basics of Image Processing
• Principle of Image Processing
• Inspection Functions
• Feature Description

• Explanation of Functions
• Peripheral Equipment
• Area Judgment by Binarization
• Position Detection
• Dimension Measurement

• Ask the Experts
• Price Info
• Free Trial Unit
• Downloads
• Contact Us

Home > Technology > Basics of Image Processing

Ask the ExpertPrice InfoTrial UnitCatalogPrint Page

Machine Vision Technology

---

Basics of Image Processing

Image output from CCD camera

As shown on the right, a target pattern focused on the image pickup element (CCD) is stored as an electric charge proportional to the brightness (incident light quantity) of each part of the image. This data is sequentially read (scanned) from its edge, and then converted into an image signal whose level (brightness/darkness information) changes with time. The image signal is output as a video signal together with a vertical synchronization signal (which defines the starting point of a screen), and a horizontal synchronization signal (which defines the starting point of a scanning line). These synchronizing signals are required to reconstruct the image.

Image scanning

Scanning transfers an image captured by a CCD to a CRT by tracing sequential horizontal lines starting at the upper left-hand corner of the screen and working down. Each line is called a scan line. There are two NTSC (National Television System Committee) scanning standards: noninterlaced and interlaced.

Non-interlaced scanning

Scans from scan line 1 to scan line 525 sequentially and then begins again at scan line 1, 30 times per second, as shown in Fig. 1.

Interlaced scanning

Scans the odd-numbered scan lines (odd field) first, then the even-numbered scan lines (even field), as shown in Fig. 2. The picture area is scanned 60 times per second but because only half of the lines are scanned at one time, the total screen area, called the frame, is scanned at a frame rate of 30 frames per second.

The CV Series conforms to the NTSC standard for non-interlaced scanning.

Binary digital conversion

A video signal sent from a camera is an analog signal. To use the video signal for various differentiations and measurements, the analog signal must be converted into a digital signal. To convert from analog to digital, a threshold level is set for the video signal. The areas brighter than the threshold level are defined as "white" and the areas darker than the binary level are defined as "black." Digital signals corresponding to a white pixel are defined as "1" (= HI), and those corresponding to a black pixel are defined as "0" (= LO).

Pixels

The video signal sent from a camera includes brightness/ darkness information that changes with time. However, the time (position on the screen) cannot be determined with this signal. When a scanning line is divided up based on a clock pulse with a specified frequency as shown in the previous column, the horizontal position on the screen can be determined. Since the vertical position is originally determined by the scanning line number, the screen is divided like a grid. Each element in the grid is called a pixel. A target image is recognized as a combination of white and black pixels. All processes are performed based on pixels.

Grayscale processing

In addition to the binary conversion method, the gray processing method is also used in image processing devices. The CV Series employs the gray processing method, which is based on the brightness graduation data of the image captured by the camera. The binary conversion method recognizes only white or black (1 or 0) data. The gray processing method divides the brightness graduation into 8 bits (256 levels), and obtains a differentiation result based on all the data. Therefore this method offers a much improved and more accurate detection.

Color processing (Color binary conversion by color extraction)

The color video signal from the camera is converted into RGB digital data by the A/D conversion of the image. This data is used for differential operation to obtain data of R- (minus) G, B-G and R-B from the received RGB data. These six color information parameters are used to check the matching degree with the color specified. This is achieved by setting the range on the screen and then extracting the color that matches the one previously specified. Then, each pixel is binary-converted into an extracted pixel or an unextracted one. This differential operation process ensures stable extraction even for dark colors and high-speed processing.

Color Shade-Scale processing

Color information data is divided into 256 levels.

Based on the extracted color, colors are divided into 256 levels. The extracted color is specified as level 255, and other colors with a greater difference in color shade data from the extracted color are specified as closer to level 0. Unlike color binary conversion, color Shade-Scale processing utilizes 256-level shade data, and therefore this processing ensures stable detection even when the color of a target varies due to individual differences. Like color binary conversion, the six parameters are used for internal operation.

Example: The case of red (R) is explained below. (The same explanation applies to the other parameters.)

Features of color Shade-Scale processing

1. Even when the ambient brightness changes or the color of a target varies due to individual differences, color Shade-Scale processing ensures more stable detection than binary conversion.
2. Since color Shade-Scale processing utilizes 256-level data, it is more effective for position measurement than binary conversion. (Sub-pixel processing is possible.)

Basics of filter processing

The term "filter processing" refers to modifications applied to a raw captured image in order to enhance specific features on the image. Typically, the filter will change the properties of a single pixel based on information gathered from surrounding pixels (using the 3 x 3 area around the modified pixel). Three images (0 to 255 tones), each in vertical and horizontal directions, are filter-processed. 240,000 filter processing is necessary for filter processing of 512 x 480 pixels.

Expand filter

An Expand filter will replace the center pixel with the brightest of the surrounding 3 x 3 pixel grid. This processing helps remove the dark-colored noise components. (Refer to the diagram below.)

Shrink filter

On the contrary, a Shrink filter replaces the central pixel with the darkest of the surrounding nine pixels. This processing helps remove the light-colored noise components. (Refer to the diagram below.) Minor defects such as dust or dirt can either be ignored using the Expand filter, or enhanced using the Shrink filter.