New method for detecting outdoor full-color LED display screens

LED (Light Emitting Diode) is one of the fastest growing industries in the world today. The characteristics of LED high brightness, low energy consumption and long life make LED display screens have obvious advantages in the field of outdoor flat panel displays. However, the differences in optical and electrical characteristics between LEDs usually cause inconsistent brightness and chromaticity of LED display screens, which in turn destroys the white balance of the display screen, reduces the display quality, and in severe cases, causes problems such as screen distortion and mosaics. In solving this problem, previous studies have mainly focused on the differences in the optoelectronic characteristics of a single LED, with the aim of finding a suitable compensation curve for RGB (red, green, and blue) three-primary color LEDs to correct their drive control parameters to improve the display effect. This type of detection and correction scheme can better solve serious problems such as screen distortion and mosaics. However, even LEDs of the same primary color and the same batch have characteristic differences, and LED full-color display screens contain many LED pixels. Various problems are inevitable during the production and manufacturing process, which will cause a certain LED pixel to not light up, or produce brightness and chromaticity differences. Therefore, this type of detection scheme has a poor correction effect on a single LED pixel, and the display effect is limited. As a compensation solution, manual visual inspection can only detect individual LED pixels with obvious differences, and requires high debugging experience from the inspectors; at the same time, the high brightness of the LED also increases the workload of the inspectors, resulting in low detection efficiency.

Therefore, this paper starts from the overall outdoor full-color LED display screen and uses digital image processing methods to quickly detect each LED pixel on the display screen, with the aim of improving the detection speed and accuracy, thereby improving the display effect of the outdoor full-color LED display screen.

1 Detection principle

As shown in Figure 1, the computer processes the display image of the LED display screen collected by the CCD (Charge Coupled Devices) sensor through the image acquisition/control module. The processing process mainly includes two parts: positioning of LED pixels and rapid detection of brightness and color.

1.1 Positioning of LED pixels

To determine the location of LED pixels, the collected LED display screen image must first be binarized. From the histogram-based image threshold segmentation method, we can know that the image is composed of one or more objects and backgrounds with different gray levels that can be separated. According to this principle, there will be multiple peaks in the histogram of the image, each peak corresponds to an object or background. To separate different objects, the valley point can be used as the threshold to divide adjacent peaks.

Due to the dot matrix characteristics of the LED display screen, it is found in the actual detection that the collected image (as shown in Figure 2 (a) and its grayscale histogram (as shown in Figure 2 (b)) have a very obvious bimodal distribution feature. For this kind of situation, the maximum variance threshold method of formula (1) is used to automatically select the segmentation threshold, which is not only effective but also fast.

Figure 1 Schematic diagram of the detection system

Where T represents the segmentation threshold, w0 and w1 represent the proportion of pixels with grayscale values ​​less than T and greater than T in the image, respectively, and 0 and 1 represent the grayscale average of the entire image, the grayscale average of the part of the image with grayscale values ​​less than T, and the grayscale average of the part of the image with grayscale values ​​greater than T, respectively.

The threshold T calculated by formula (1) is used to binarize the grayscale image of Figure 2 (a) to obtain Figure 2 (c). Then, Figure 2 (c) is projected horizontally and vertically respectively to calculate the position of the LED pixel on the display screen.

2 (a) Collected blue image 2 (b) Grayscale histogram 2 (c) Binarized image

Figure 2 Positioning processing results

1.2 Rapid detection of LED pixel brightness and chromaticity

By referring to the YUV color model (Y represents brightness, U and V are two components that constitute color) successfully used in PAL (Phase Alternating Line) television systems, the RGB color model used in the image is converted into the color model of formula (2), which can conveniently and quickly calculate the relative brightness value of each pixel.

According to the additive color principle in colorimetry, the outdoor full-color LED display screen is composed of RGB three-primary color LEDs to form each pixel on the display screen. By controlling the luminous intensity of a certain primary color LED in each pixel, various colors can be matched to display a colorful color image on the display screen. In the CIE (International Commission on Illumination) RG chromaticity diagram, the chromaticity coordinates reflect the relative proportion of each of the three primary colors in the total amount of the three stimulus values. A set of chromaticity coordinates represents the common characteristics of those colors with the same hue and saturation but different brightness.

Each pixel on the LED display can always find a corresponding area in the image to be tested. Therefore, the chromaticity of the pixel can be determined by the RGB value in the image data in its corresponding area, and the calculation formula is as shown in formula (3).

Assuming the measured brightness value of the LED pixel is Y1 and the chromaticity coordinates are (r1, g1), by analyzing the discreteness of Y1 and (ri, g1), the LED pixels with inconsistent brightness and chromaticity on the LED display can be determined.

To verify the effectiveness of the detection method, this paper uses the AvaSpec-2048 micro-spectrometer to conduct a comparative test of the brightness and chromaticity of the unit modules of the same outdoor full-color LED display. In order to reduce the amount of calculation and facilitate debugging, this paper uses the CIErg chromaticity coordinate system, which is different from the internationally used CIExy chromaticity coordinate system used by the spectrometer. Therefore, the chromaticity coordinates must be converted during the test, as shown in formula (4).

2. Processing results and analysis

This paper uses CCD image sensors to collect images and performs algorithm testing on the LED pixels in the unit module of a three-in-one surface-mount outdoor full-color LED display.

Taking blue as an example, Figure 2 (a) is a blue image of a three-in-one surface mount unit module captured by a CCD image sensor. To better verify the effectiveness of the detection method, this paper shields some pixels of the LED display unit module to form the black part in Figure 2 (a).

Figure 3 MacAdam color wide capacity ellipse

Since LED is a self-luminous body, and the luminous intensity is proportional to the driving current provided to it within a certain range, the brightness difference can be minimized by reasonably controlling the driving current during the design, manufacture and debugging of the driving circuit. The average value is used as the standard value, which should be less than 15% to 20%. Therefore, in order to facilitate the subsequent brightness correction, the experiment locates and counts the LED pixels that deviate from the overall brightness average value by more than 5%, so as to control the brightness difference of these pixels with large deviations within 10%. When performing chromaticity detection, this paper refers to the method of quantifying color tolerance by MacAdam (DLMacAdam) (as shown in Figure 3), and counts the chromaticity coordinates of each LED pixel, finds the geometric center of these chromaticity coordinates, and records 3-5% of the LED pixels whose Euclidean distance from the geometric center is greater than d0 (the d0 value is different for different colors), as shown in formula (5).

Table 1 shows the detection results (taking blue as an example), where the brightness value Y1 is the relative brightness, which is proportional to the maximum brightness 255; the chromaticity coordinates are (r1, g1).

Table 1 Test result statistics (blue)

The same unit module was tested by AvaSpec-2048 micro-spectrometer, and the test results are shown in Table 2. The comparison shows that the detection method used in this paper is effective and feasible, and has fast detection speed and high accuracy.

Table 2 AvaSpec-2048 micro-spectrometer test results (blue)

3 Conclusion

This paper uses CCD image sensors and digital image processing technology to evaluate the brightness and chromaticity uniformity of outdoor full-color LED display screens. A new rapid detection method is proposed, which better ensures the consistency of the display effect of each LED pixel on the display screen, and provides a reference basis for quantitative debugging for subsequent brightness and chromaticity correction work, which can greatly improve the detection efficiency and display quality of outdoor full-color LED display screens. The next step will be to continue to study the influence of ambient light on brightness and chromaticity detection and how to overcome it, as well as the research on the brightness and chromaticity automatic correction drive circuit, and finally realize the accurate detection and correction of the brightness and chromaticity values ​​of each LED pixel on the outdoor full-color display screen.