tft lcd image converter made in china

This application allows you to create bitmaps and fonts, and transform them to “C” source format for embedded applications. The transformation of the images to the source code is made by using templates. Therefore, by modifying the templates, you can change the format of the output within certain limits.

The chip includes a touch controller for sensing touch feedback from a TFT display with a resistive touch panel as well as a sound synthesizer and audio playback with a PWM output to drive a speaker.

In this article, you will learn how to use TFT LCDs by Arduino boards. From basic commands to professional designs and technics are all explained here.

There are several components to achieve this. LEDs,  7-segments, Character and Graphic displays, and full-color TFT LCDs. The right component for your projects depends on the amount of data to be displayed, type of user interaction, and processor capacity.

TFT LCD is a variant of a liquid-crystal display (LCD) that uses thin-film-transistor (TFT) technology to improve image qualities such as addressability and contrast. A TFT LCD is an active matrix LCD, in contrast to passive matrix LCDs or simple, direct-driven LCDs with a few segments.

In Arduino-based projects, the processor frequency is low. So it is not possible to display complex, high definition images and high-speed motions. Therefore, full-color TFT LCDs can only be used to display simple data and commands.

There are several components to achieve this. LEDs,  7-segments, Character and Graphic displays, and full-color TFT LCDs. The right component for your projects depends on the amount of data to be displayed, type of user interaction, and processor capacity.

TFT LCD is a variant of a liquid-crystal display (LCD) that uses thin-film-transistor (TFT) technology to improve image qualities such as addressability and contrast. A TFT LCD is an active matrix LCD, in contrast to passive matrix LCDs or simple, direct-driven LCDs with a few segments.

In Arduino-based projects, the processor frequency is low. So it is not possible to display complex, high definition images and high-speed motions. Therefore, full-color TFT LCDs can only be used to display simple data and commands.

Size of displays affects your project parameters. Bigger Display is not always better. if you want to display high-resolution images and signs, you should choose a big size display with higher resolution. But it decreases the speed of your processing, needs more space and also needs more current to run.

After choosing the right display, It’s time to choose the right controller. If you want to display characters, tests, numbers and static images and the speed of display is not important, the Atmega328 Arduino boards (such as Arduino UNO) are a proper choice. If the size of your code is big, The UNO board may not be enough. You can use Arduino Mega2560 instead. And if you want to show high resolution images and motions with high speed, you should use the ARM core Arduino boards such as Arduino DUE.

In electronics/computer hardware a display driver is usually a semiconductor integrated circuit (but may alternatively comprise a state machine made of discrete logic and other components) which provides an interface function between a microprocessor, microcontroller, ASIC or general-purpose peripheral interface and a particular type of display device, e.g. LCD, LED, OLED, ePaper, CRT, Vacuum fluorescent or Nixie.

The display driver will typically accept commands and data using an industry-standard general-purpose serial or parallel interface, such as TTL, CMOS, RS232, SPI, I2C, etc. and generate signals with suitable voltage, current, timing and demultiplexing to make the display show the desired text or image.

The LCDs manufacturers use different drivers in their products. Some of them are more popular and some of them are very unknown. To run your display easily, you should use Arduino LCDs libraries and add them to your code. Otherwise running the display may be very difficult. There are many free libraries you can find on the internet but the important point about the libraries is their compatibility with the LCD’s driver. The driver of your LCD must be known by your library. In this article, we use the Adafruit GFX library and MCUFRIEND KBV library and example codes. You can download them from the following links.

First you should convert your image to hex code. Download the software from the following link. if you don’t want to change the settings of the software, you must invert the color of the image and make the image horizontally mirrored and rotate it 90 degrees counterclockwise. Now add it to the software and convert it. Open the exported file and copy the hex code to Arduino IDE. x and y are locations of the image. sx and sy are sizes of image. you can change the color of the image in the last input.

Upload your image and download the converted file that the UTFT libraries can process. Now copy the hex code to Arduino IDE. x and y are locations of the image. sx and sy are size of the image.

In this template, We converted a .jpg image to .c file and added to the code, wrote a string and used the fade code to display. Then we used scroll code to move the screen left. Download the .h file and add it to the folder of the Arduino sketch.

In this template, We used sin(); and cos(); functions to draw Arcs with our desired thickness and displayed number by text printing function. Then we converted an image to hex code and added them to the code and displayed the image by bitmap function. Then we used draw lines function to change the style of the image. Download the .h file and add it to the folder of the Arduino sketch.

In this template, We added a converted image to code and then used two black and white arcs to create the pointer of volumes.  Download the .h file and add it to the folder of the Arduino sketch.

In this template, We added a converted image and use the arc and print function to create this gauge.  Download the .h file and add it to folder of the Arduino sketch.

while (a < b) { Serial.println(a); j = 80 * (sin(PI * a / 2000)); i = 80 * (cos(PI * a / 2000)); j2 = 50 * (sin(PI * a / 2000)); i2 = 50 * (cos(PI * a / 2000)); tft.drawLine(i2 + 235, j2 + 169, i + 235, j + 169, tft.color565(0, 255, 255)); tft.fillRect(200, 153, 75, 33, 0x0000); tft.setTextSize(3); tft.setTextColor(0xffff); if ((a/20)>99)

while (b < a) { j = 80 * (sin(PI * a / 2000)); i = 80 * (cos(PI * a / 2000)); j2 = 50 * (sin(PI * a / 2000)); i2 = 50 * (cos(PI * a / 2000)); tft.drawLine(i2 + 235, j2 + 169, i + 235, j + 169, tft.color565(0, 0, 0)); tft.fillRect(200, 153, 75, 33, 0x0000); tft.setTextSize(3); tft.setTextColor(0xffff); if ((a/20)>99)

In this template, We display simple images one after each other very fast by bitmap function. So you can make your animation by this trick.  Download the .h file and add it to folder of the Arduino sketch.

In this template, We just display some images by RGBbitmap and bitmap functions. Just make a code for touchscreen and use this template.  Download the .h file and add it to folder of the Arduino sketch.

In this Arduino touch screen tutorial we will learn how to use TFT LCD Touch Screen with Arduino. You can watch the following video or read the written tutorial below.

As an example I am using a 3.2” TFT Touch Screen in a combination with a TFT LCD Arduino Mega Shield. We need a shield because the TFT Touch screen works at 3.3V and the Arduino Mega outputs are 5 V. For the first example I have the HC-SR04 ultrasonic sensor, then for the second example an RGB LED with three resistors and a push button for the game example. Also I had to make a custom made pin header like this, by soldering pin headers and bend on of them so I could insert them in between the Arduino Board and the TFT Shield.

Here’s the circuit schematic. We will use the GND pin, the digital pins from 8 to 13, as well as the pin number 14. As the 5V pins are already used by the TFT Screen I will use the pin number 13 as VCC, by setting it right away high in the setup section of code.

I will use the UTFT and URTouch libraries made by Henning Karlsen. Here I would like to say thanks to him for the incredible work he has done. The libraries enable really easy use of the TFT Screens, and they work with many different TFT screens sizes, shields and controllers. You can download these libraries from his website, RinkyDinkElectronics.com and also find a lot of demo examples and detailed documentation of how to use them.

After we include the libraries we need to create UTFT and URTouch objects. The parameters of these objects depends on the model of the TFT Screen and Shield and these details can be also found in the documentation of the libraries.

So now I will explain how we can make the home screen of the program. With the setBackColor() function we need to set the background color of the text, black one in our case. Then we need to set the color to white, set the big font and using the print() function, we will print the string “Arduino TFT Tutorial” at the center of the screen and 10 pixels  down the Y – Axis of the screen. Next we will set the color to red and draw the red line below the text. After that we need to set the color back to white, and print the two other strings, “by HowToMechatronics.com” using the small font and “Select Example” using the big font.

Many times to make any user interface or nicely display content, we use icons/images. In this Instructable, you will be displaying icons or logos or images on your TFT screen from Arduino with using ATmega (microcontroller used in Arduino) Flash memory. It does not require any SD Card to store bitmap images or USB connection to send image data. We will convert images from any image format like .bmp, .jpg, .jpeg, .png to its hexadecimal equivalent to be stored in flash memory of arduino mega (ATmega2560).

All microcontroller has Flash memory, where the codes are stored permanently. Arduino Mega has comparatively good amount of Flash memory, ie 256 KB of which 8 KB used by bootloader. We will be doing two things:Monochrome icons/images: The icons or images will be displayed with single color, but takes very less memory. Just 1bit for one pixel.

Colored icons/images: It depends on the TFT screen used, for eg. 1.8" SPI TFT with ST7735 driver has 16bit color. Images or icons will just look like your phone screen, but it takes lots of space. it takes 16bits (2bytes) for each pixel (16times more!!).

It requires a TFT screen compatible with arduino, few jumper cables (dupont wires), breadboard and is recommended to use 3.3V -5V level shifters (but it works without it also :P ). But we have used evive . It has all the things required to do this without any additional wiring!! Hence it helps in avoiding the repetitive task for bread-boarding. evive uses the most commonly used 1.8" SPI based TFT (ST7735R driver) having 160px by 128px along with Arduino Mega 2560 R3. Also has internal logic level shifters for ideal usage.

https://sourceforge.net/projects/lcd-image-convert...This tool has all the options for large varieties of screens available. You can even draw your own icon!!.

Also we may need to use some image resizing tool as most of the images available on internet are of very large size as compared to hoscreen. Option for Image Resizer:

If you are using the tool mentioned in last step, Please look at the images. It has lots of options to resize image for our usage. We can easily enter the value of "height" or "width" in pixels!.

Once you have the image ready, next step is to convert the image to some form of numbers as actually all images are represented by array/matrix of numbers. Since we are not going to use SD card to save images or logos or icons as its irritating everytime to have a micro SD card for this purpose, we will now convert images to hexadecimal. Then we will store it in Arduino Flash Memory.

Using the LCD_Image_Converter tool, we will get the image in hexadecimal form.Load the image usign File->Open->"SelectUrImage" If you want to edit image, use the editing tools.

Copy all the numbers!! (Here each pixel is stored in its binary form as image is monochrome. If the pixel is filled, then it will be 1 or else it will be 0)

Other option is to go for colored images (remember that it takes lot of Arduino Flash memory). Based on the TFT screen you will have to select some options like color format (1.8" SPI TFT SR7735R uses 16 bit colors: R5G6B5)Load the image usign File->Open->"SelectUrImage" If you want to edit image, use the editing tools.

Case 2: Since there is no direct function in Adafruit GFX library, we will have to write our own code for this. We will have to traverse pixel by pixel to draw image/icon.

LCD, or Liquid Crystal Displays, are great choices for many applications. They aren’t that power-hungry, they are available in monochrome or full-color models, and they are available in all shapes and sizes.

Waveshare actually has several round LCD modules, I chose the 1.28-inch model as it was readily available on Amazon. You could probably perform the same experiments using a different module, although you may require a different driver.

This display can be used for the experiments we will be doing with the ESP32, as that is a 3.3-volt logic microcontroller. You would need to use a voltage level converter if you wanted to use one of these with an Arduino Uno.

Open the Arduino folder. Inside you’ll find quite a few folders, one for each display size that Waveshare supports. As I’m using the 1.28-inch model, I selected theLCD_1inch28folder.

Once you do that, you can open your Arduino IDE and then navigate to that folder. Inside the folder, there is a sketch file namedLCD_1inch28.inowhich you will want to open.

You can see from the code that after loading some libraries we initialize the display, set its backlight level (you can use PWM on the BL pin to set the level), and paint a new image. We then proceed to draw lines and strings onto the display.

Unfortunately, Waveshare doesn’t offer documentation for this, but you can gather quite a bit of information by reading theLCD_Driver.cppfile, where the functions are somewhat documented.

The TFT_eSPI library is ideal for this, and several other, displays. You can install it through your Arduino IDE Library Manager, just search for “TFT_eSPI”.

The Animated Eyes sketch can be found within the sample files for the TFT_eSPI library, under the “generic” folder.  Assuming that you have wired up the second GC9A01 display, you’ll want to use theAnimated_Eyes_2sketch.

The GC9A01 LCD module is a 1.28-inch round display that is useful for instrumentation and other similar projects. Today we will learn how to use this display with an Arduino Uno and an ESP32.

A thin-film-transistor liquid-crystal display (TFT LCD) is a variant of a liquid-crystal display that uses thin-film-transistor technologyactive matrix LCD, in contrast to passive matrix LCDs or simple, direct-driven (i.e. with segments directly connected to electronics outside the LCD) LCDs with a few segments.

In February 1957, John Wallmark of RCA filed a patent for a thin film MOSFET. Paul K. Weimer, also of RCA implemented Wallmark"s ideas and developed the thin-film transistor (TFT) in 1962, a type of MOSFET distinct from the standard bulk MOSFET. It was made with thin films of cadmium selenide and cadmium sulfide. The idea of a TFT-based liquid-crystal display (LCD) was conceived by Bernard Lechner of RCA Laboratories in 1968. In 1971, Lechner, F. J. Marlowe, E. O. Nester and J. Tults demonstrated a 2-by-18 matrix display driven by a hybrid circuit using the dynamic scattering mode of LCDs.T. Peter Brody, J. A. Asars and G. D. Dixon at Westinghouse Research Laboratories developed a CdSe (cadmium selenide) TFT, which they used to demonstrate the first CdSe thin-film-transistor liquid-crystal display (TFT LCD).active-matrix liquid-crystal display (AM LCD) using CdSe TFTs in 1974, and then Brody coined the term "active matrix" in 1975.high-resolution and high-quality electronic visual display devices use TFT-based active matrix displays.

The liquid crystal displays used in calculators and other devices with similarly simple displays have direct-driven image elements, and therefore a voltage can be easily applied across just one segment of these types of displays without interfering with the other segments. This would be impractical for a large display, because it would have a large number of (color) picture elements (pixels), and thus it would require millions of connections, both top and bottom for each one of the three colors (red, green and blue) of every pixel. To avoid this issue, the pixels are addressed in rows and columns, reducing the connection count from millions down to thousands. The column and row wires attach to transistor switches, one for each pixel. The one-way current passing characteristic of the transistor prevents the charge that is being applied to each pixel from being drained between refreshes to a display"s image. Each pixel is a small capacitor with a layer of insulating liquid crystal sandwiched between transparent conductive ITO layers.

The circuit layout process of a TFT-LCD is very similar to that of semiconductor products. However, rather than fabricating the transistors from silicon, that is formed into a crystalline silicon wafer, they are made from a thin film of amorphous silicon that is deposited on a glass panel. The silicon layer for TFT-LCDs is typically deposited using the PECVD process.

Polycrystalline silicon is sometimes used in displays requiring higher TFT performance. Examples include small high-resolution displays such as those found in projectors or viewfinders. Amorphous silicon-based TFTs are by far the most common, due to their lower production cost, whereas polycrystalline silicon TFTs are more costly and much more difficult to produce.

The twisted nematic display is one of the oldest and frequently cheapest kind of LCD display technologies available. TN displays benefit from fast pixel response times and less smearing than other LCD display technology, but suffer from poor color reproduction and limited viewing angles, especially in the vertical direction. Colors will shift, potentially to the point of completely inverting, when viewed at an angle that is not perpendicular to the display. Modern, high end consumer products have developed methods to overcome the technology"s shortcomings, such as RTC (Response Time Compensation / Overdrive) technologies. Modern TN displays can look significantly better than older TN displays from decades earlier, but overall TN has inferior viewing angles and poor color in comparison to other technology.

The transmittance of a pixel of an LCD panel typically does not change linearly with the applied voltage,sRGB standard for computer monitors requires a specific nonlinear dependence of the amount of emitted light as a function of the RGB value.

Less expensive PVA panels often use dithering and FRC, whereas super-PVA (S-PVA) panels all use at least 8 bits per color component and do not use color simulation methods.BRAVIA LCD TVs offer 10-bit and xvYCC color support, for example, the Bravia X4500 series. S-PVA also offers fast response times using modern RTC technologies.

A technology developed by Samsung is Super PLS, which bears similarities to IPS panels, has wider viewing angles, better image quality, increased brightness, and lower production costs. PLS technology debuted in the PC display market with the release of the Samsung S27A850 and S24A850 monitors in September 2011.

TFT dual-transistor pixel or cell technology is a reflective-display technology for use in very-low-power-consumption applications such as electronic shelf labels (ESL), digital watches, or metering. DTP involves adding a secondary transistor gate in the single TFT cell to maintain the display of a pixel during a period of 1s without loss of image or without degrading the TFT transistors over time. By slowing the refresh rate of the standard frequency from 60 Hz to 1 Hz, DTP claims to increase the power efficiency by multiple orders of magnitude.

Due to the very high cost of building TFT factories, there are few major OEM panel vendors for large display panels. The glass panel suppliers are as follows:

External consumer display devices like a TFT LCD feature one or more analog VGA, DVI, HDMI, or DisplayPort interface, with many featuring a selection of these interfaces. Inside external display devices there is a controller board that will convert the video signal using color mapping and image scaling usually employing the discrete cosine transform (DCT) in order to convert any video source like CVBS, VGA, DVI, HDMI, etc. into digital RGB at the native resolution of the display panel. In a laptop the graphics chip will directly produce a signal suitable for connection to the built-in TFT display. A control mechanism for the backlight is usually included on the same controller board.

The low level interface of STN, DSTN, or TFT display panels use either single ended TTL 5 V signal for older displays or TTL 3.3 V for slightly newer displays that transmits the pixel clock, horizontal sync, vertical sync, digital red, digital green, digital blue in parallel. Some models (for example the AT070TN92) also feature input/display enable, horizontal scan direction and vertical scan direction signals.

New and large (>15") TFT displays often use LVDS signaling that transmits the same contents as the parallel interface (Hsync, Vsync, RGB) but will put control and RGB bits into a number of serial transmission lines synchronized to a clock whose rate is equal to the pixel rate. LVDS transmits seven bits per clock per data line, with six bits being data and one bit used to signal if the other six bits need to be inverted in order to maintain DC balance. Low-cost TFT displays often have three data lines and therefore only directly support 18 bits per pixel. Upscale displays have four or five data lines to support 24 bits per pixel (truecolor) or 30 bits per pixel respectively. Panel manufacturers are slowly replacing LVDS with Internal DisplayPort and Embedded DisplayPort, which allow sixfold reduction of the number of differential pairs.

Kawamoto, H. (2012). "The Inventors of TFT Active-Matrix LCD Receive the 2011 IEEE Nishizawa Medal". Journal of Display Technology. 8 (1): 3–4. Bibcode:2012JDisT...8....3K. doi:10.1109/JDT.2011.2177740. ISSN 1551-319X.

K. H. Lee; H. Y. Kim; K. H. Park; S. J. Jang; I. C. Park & J. Y. Lee (June 2006). "A Novel Outdoor Readability of Portable TFT-LCD with AFFS Technology". SID Symposium Digest of Technical Papers. AIP. 37 (1): 1079–82. doi:10.1889/1.2433159. S2CID 129569963.

As a 2.4inch TFT display module with a resolution of 240 * 320, it uses the SPI interface for communication. LCD has an internal controller with basic functions, which can be used to draw points, lines, circles, and rectangles, and can display English, Chinese as well as pictures.

The 2.4inch LCD uses the PH2.0 8PIN interface, which can be connected to the Raspberry Pi according to the above table: (Please connect according to the pin definition table. The color of the wiring in the picture is for reference only, and the actual color shall prevail.)

The LCD supports 12-bit, 16-bit, and 18-bit input color formats per pixel, namely RGB444, RGB565, and RGB666 three color formats, this demo uses RGB565 color format, which is also a commonly used RGB format.

For most LCD controllers, the communication mode of the controller can be configured, usually with an 8080 parallel interface, three-wire SPI, four-wire SPI, and other communication methods. This LCD uses a four-wire SPI communication interface, which can greatly save the GPIO port, and the communication speed will be faster.

Select image buffer: The purpose of the selection is that you can create multiple image attributes, there can be multiple images buffer, you can select each image you create.

Mirror: indicates the image mirroring mode. MIRROR_NONE, MIRROR_HORIZONTAL, MIRROR_VERTICAL, MIRROR_ORIGIN correspond to no mirror, horizontal mirror, vertical mirror, and image center mirror respectively.

The fill color of a certain window in the image buffer: the image buffer part of the window filled with a certain color, usually used to fresh the screen into blank, often used for time display, fresh the last second of the screen.

Draw rectangle: In the image buffer, draw a rectangle from (Xstart, Ystart) to (Xend, Yend), you can choose the color, the width of the line, whether to fill the inside of the rectangle.

Draw circle: In the image buffer, draw a circle of Radius with (X_Center Y_Center) as the center. You can choose the color, the width of the line, and whether to fill the inside of the circle.

Write Ascii character: In the image buffer, use (Xstart Ystart) as the left vertex, write an Ascii character, you can select Ascii visual character library, font foreground color, font background color.

Write English string: In the image buffer, use (Xstart Ystart) as the left vertex, write a string of English characters, you can choose Ascii visual character library, font foreground color, font background color.

Write Chinese string: in the image buffer, use (Xstart Ystart) as the left vertex, write a string of Chinese characters, you can choose character font, font foreground color, font background color of the GB2312 encoding

Write numbers: In the image buffer,use (Xstart Ystart) as the left vertex, write a string of numbers, you can choose Ascii visual character library, font foreground color, font background color.

Display time: in the image buffer,use (Xstart Ystart) as the left vertex, display time,you can choose Ascii visual character font, font foreground color, font background color.;

2. The module_init() function is automatically called in the INIT () initializer on the LCD, but the module_exit() function needs to be called by itself

Python has an image library PIL official library link, it do not need to write code from the logical layer like C, can directly call to the image library for image processing. The following will take 1.54inch LCD as an example, we provide a brief description for the demo.

The first parameter defines the color depth of the image, which is defined as "1" to indicate the bitmap of one-bit depth. The second parameter is a tuple that defines the width and height of the image. The third parameter defines the default color of the buffer, which is defined as "WHITE".

The demo is developed based on the HAL library. Download the demo, find the STM32 program file directory, and open the LCD_demo.uvprojx in the STM32\STM32F103RBT6\MDK-ARM directory to check the program.

Select image buffer:the purpose of the selection is that you can create multiple image attributes, image buffer can exist multiple, you can select each image you create.

Mirror: indicates the image mirroring mode. MIRROR_NONE, MIRROR_HORIZONTAL, MIRROR_VERTICAL, MIRROR_ORIGIN correspond to no mirror, horizontal mirror, vertical mirror, and about image center mirror respectively.

Image buffer part of the window filling color: the image buffer part of the window filled with a certain color, generally as a window whitewashing function, often used for time display, whitewashing on a second

Draw rectangle: In the image buffer, draw a rectangle from (Xstart, Ystart) to (Xend, Yend), you can choose the color, the width of the line, whether to fill the inside of the rectangle.

Draw circle: In the image buffer, draw a circle of Radius with (X_Center Y_Center) as the center. You can choose the color, the width of the line, and whether to fill the inside of the circle.

Write Ascii character: In the image buffer, at (Xstart Ystart) as the left vertex, write an Ascii character, you can select Ascii visual character library, font foreground color, font background color.

Write English string: In the image buffer, use (Xstart Ystart) as the left vertex, write a string of English characters, can choose Ascii visual character library, font foreground color, font background color.

Write Chinese string: in the image buffer, use (Xstart Ystart) as the left vertex, write a string of Chinese characters, you can choose GB2312 encoding character font, font foreground color, font background color.

Write numbers: In the image buffer,use (Xstart Ystart) as the left vertex, write a string of numbers, you can choose Ascii visual character library, font foreground color, font background color.

Display time: in the image buffer,use (Xstart Ystart) as the left vertex, display time,you can choose Ascii visual character font, font foreground color, font background color.

image.cpp(.h): is the image data, which can convert any BMP image into a 16-bit true color image array through Img2Lcd (downloadable in the development data).

Select image buffer:the purpose of the selection is that you can create multiple image attributes, image buffer can exist multiple, you can select each image you create.

Mirror: indicates the image mirroring mode. MIRROR_NONE, MIRROR_HORIZONTAL, MIRROR_VERTICAL, MIRROR_ORIGIN correspond to no mirror, horizontal mirror, vertical mirror, and about image center mirror respectively.

Draw rectangle: In the image buffer, draw a rectangle from (Xstart, Ystart) to (Xend, Yend), you can choose the color, the width of the line, whether to fill the inside of the rectangle.

Draw circle: In the image buffer, draw a circle of Radius with (X_Center Y_Center) as the center. You can choose the color, the width of the line, and whether to fill the inside of the circle.

Write Ascii character: In the image buffer, at (Xstart Ystart) as the left vertex, write an Ascii character, you can select Ascii visual character library, font foreground color, font background color.

Write English string: In the image buffer, use (Xstart Ystart) as the left vertex, write a string of English characters, can choose Ascii visual character library, font foreground color, font background color.

Write Chinese string: in the image buffer, use (Xstart Ystart) as the left vertex, write a string of Chinese characters, you can choose GB2312 encoding character font, font foreground color, font background color.

Write numbers: In the image buffer,use (Xstart Ystart) as the left vertex, write a string of numbers, you can choose Ascii visual character library, font foreground color, font background color.

Display time: in the image buffer,use (Xstart Ystart) as the left vertex, display time,you can choose Ascii visual character font, font foreground color, font background color.

According to real LCD manufacturing conditions, the number of normal LCD panels exceeds greatly the number of defective LCD panels. Therefore, the normal PRs greatly outnumber the defective PRs. As a result, the collected data set for training would be imbalanced if a two-class classification approach is adopted, the SVM by Vapnik [4] for example, the class imbalance problem occurs.

In practice, in addition to the class imbalance problem, the LCD defect detection also suffers from another critical problem resulting from the absence of negative information. To facilitate the following problem description, the normal PR class and the defective PR class are defined as the positive class and negative class, respectively.

The main difference between a normal PR and a defective PR is that their appearances are apparently different, as can be observed from Figure 4. The color (or gray level) of a normal PR is nearly uniform, implying that the variation of the gray-level distribution of normal PRs is very small. On the contrary, the surfaces of defective PR not only contain various kinds of textures, but also vary greatly in color, implying that the variation of the true distribution for negative class in the data space is very large. Collecting a set of positive training data that can represent the true distribution of positive class is easy, because: (1) the variation of positive-class distribution is very small; and (2) most of the LCD panels are normal (the number of normal PRs is considerably large). Therefore, the positive class can be well-sampled during the data collection stage in real practice. However, representative defective PRs are difficult to obtain in practice for several reasons. For example, there are numerous types of defects in array process, more than 10 types at least. However, not all the defects would occur frequently. Some of the defects seldom appear, for example the defect caused by abnormal photo-resist coating (APRC). The defect “APRC” seldom occurs, because equipment/process engineers maintain the coating machines periodically. Even so, the coating machines might still break down occasionally. As a result, the number of available images containing the APRC defects is quite limited. But, the APRC defect has a large variation in color and texture. Unfortunately, limited APRC examples cannot stand for all kinds of APRC defects. Therefore, the collected negative training data are most likely under-sampled. Here, the “under-sampled” means that the collected negative training set cannot represent the true negative-class distribution in the data space, which is the problem of absence of negative information. Due to this problem, numerous false positive (i.e., missing defects) will be produced if a two-class classification approach (e.g., a binary SVM) is applied to the LCD defect detection, which has been evidenced by the results reported in [7]. Compared with two-class classification approach, novelty detection approach is a better choice.

Different defect images contain different numbers of defective pixel regions (PRs). The normal and defective PRs are bounded with blue and red rectangles, respectively.

Novelty detection is one-class classification [10,35], which is to solve the conventional two-class classification problems where one of the two classes is under-sampled, or only the data of one single class can be available for training [5,6,9–11,35–40]. As analyzed above, for the LCD defect detection application, the normal PRs can be well-sampled, while the defective PRs are in general undersampled. Therefore, the LCD defect detection can be treated as a typical novelty detection problem. Accordingly, one-class classification is a better solution.

To summarize, it can be seen that the LCD defect detection suffers from two problems simultaneously: one is the class imbalance problem, and the other is the problem of the absence of negative information. For the first problem, there have been many sophisticated solutions, including sampling, cost-sensitive learning, SVM-based, and one-class learning approaches. However, the only solution to the second problem is the novelty detection approach (i.e., one-class classification approach). Therefore, one-class classification would be a more appropriate approach to the LCD defect detection application.

There are several approaches for one-class classification, such as density approach (e.g., Gaussian mixture model [5]), boundary approach (e.g., SVDD [9] and one-class SVM [40]), neural network approach [6,36], and reconstruction-based approach (e.g., the kernel principal component analysis for novelty detection [35]). It has been proven in [9] that when a Gaussian kernel is used, the SVDD proposed by Tax and Duin [9] is identical to the one-class SVM proposed by Schölkopf et al. [40]. This paper focuses on the SVDD since it has been applied to the same application in the works of [7] and [10], and has shown to be effective in detecting defective PRs. However, as discussed in Section 1, generalization performance of SVDD is limited. Therefore, the intent of this paper is on proposing a method to improve generalization performance of SVDD, and applying the improved SVDD to the LCD defect detection treated as a novelty detection problem. The improved SVDD is called quasiconformal kernel SVDD (QK-SVDD). Note that the QK-SVDD and SVDD are not two independent classifiers. To obtain QK-SVDD, one has to train an SVDD first, which will be introduced in Section 2.4. In the following part of the paper, we first introduce the defect detection scheme, and then derive the proposed method in details.