In recent years, with the gradual deepening of the application of large-screen display technology in various fields, the market is no longer satisfied with a single film display, and more turned to the integrated display of the more interactive computer desktop environment. At present, mainstream desktop fusion systems on the market mostly use hardware auxiliary equipment such as splitter, which has high cost and poor performance.
CUDA (Compute Unified Device Architect-ure, CUDA) is a general-purpose parallel computing architecture launched by NVIDIA in recent years. It uses high-performance graphics GPU as hardware support, and the use of CPU + GPU hybrid computing has greatly improved Real-time processing efficiency of large-scale graphics data. The video display system designed in this paper uses the CUDA development method to realize the three aspects of computer desktop picture segmentation calculation, Bezier curve fitting, and fusion image calculation. The real-time performance is high, and the theoretical calculation of the picture data is 14 pixels, which is accurate.
1 System framework design
The essence of image processing is large-scale matrix operation, which is especially suitable for parallel processing. However, it is difficult for CPU general-purpose computing to take advantage of this feature. In contrast, GPUs have powerful computing capabilities in parallel data operations, and are particularly suitable for operations with the same operator but different operation data. When performing multiple data elements with high operation density, the delay of memory access can be ignored. The CUDA programming model uses the CPU as a host and the GPU as a coprocessor or device. There can be multiple devices in a system. In this model, the CPU and GPU work together. The CPU is responsible for highly logical transaction processing and serial computing. The GPU is focused on executing highly threaded parallel processing tasks.
This system uses the computing platform built by NVIDIA GeForce GTX470 as the operating environment. It uses the multi-head output feature of the graphics card to connect multiple projectors to form a spliced ​​screen array without additional hardware equipment. Due to the characteristics of large image data and high real-time performance of the desktop fusion display system, the software design of this system widely uses multi-CPU parallel programming technology and CUDA parallel computing technology for image processing and display of each projection device , The system will allocate a special thread to deal with. This thread will correspond to a fixed CPU and a fixed GPU computing core to ensure that the multi-projection devices are completely processed in parallel, thereby avoiding the performance degradation of other systems due to the increase in display devices and the increase in processing data. The CUDA architecture is shown in Figure 1.
In the design of this system, first set the timer. The image information of the control screen is collected regularly and saved in the public storage space, and then several threads are dynamically opened for the number of peripheral display devices to complete image data segmentation, image data fusion, and image display. Among them, the thread is initially associated with a fixed GPU computing core, and the data graphics segmentation and fusion are implemented using CUDA technology. Finally, the timer technology is also used to synchronize the display of image data in each thread.
The Bessel curve fitting method is usually used to complete the fusion of image data. The general method of this method is to first obtain the Bezier curve of each row of the target image from the control points to form a two-dimensional Bezier surface, and then fit the target image data to the Bezier curve by texture method. The image is distorted. Bezier curve is a method of defining curves and surfaces with control polygons. Its fitting interpolation formula is:
In the formula: Pi is the characteristic polygon that constitutes the curve; Bi, n (t) is the Bezier basis function and the harmonic function of the position vector of each point on the curve. The start and end points of the Bezier curve coincide with the end points of the feature polygon, and the tangent directions of the start and end points are consistent with the first and last edges of the feature polygon.
The curve has characteristics such as convex hull and symmetry. The advantage of Bezier curve is that given enough control points, it can fit a curve of any shape.
In the fitting interpolation formula of Bezier curve, the degree of the function corresponds to the number of vertices of the feature polygon. When the number of vertices of the feature polygon is 4, a cubic Bezier curve is formed. The fitting interpolation formula of cubic Bezier curve is:
OPenGL technology provides easy-to-operate Bezier curve generation and texture functions, but it cannot control hardware operations and is not efficient. In consideration of timeliness, this system does not use this method in the implementation process, but uses the CUDA technology parallel matrix operation to texture map. According to the structural characteristics of the CUDA program, when this system is processed, the Bessel curve of each line of the target image is generated by the usual method according to the control point information obtained from the human-computer interaction part. Open up the video memory and store it in the GPU, and then correspond to the CUDA program structure, and assign a GPU thread to each pixel on the target image for processing.
Observing the calculation formula above, it is found that when the coordinates of the initial control point are obtained, the points on each Bezier curve are not affected by each other, and have the coarse-grained characteristics of multithreading, so CUDA parallel calculation The timeliness has been greatly improved.
2 Performance evaluation
In the multimedia splicing system, real-time performance is the most basic and important indicator. The large screen splicing dynamic effect that we watched is produced by the rapid display of one frame after another. According to the normal human visual residual level system, to achieve a smooth display, at least 25 pictures must be processed in 1s, which means that the entire program image processing process will not exceed 40ms. The following article will analyze the timeliness of the system.
Since the image acquisition part and the processing part adopt an asynchronous method and time multiplexing, and obviously the processing part takes much longer time than the acquisition part, only the time consumption of the processing part is listed, and because the efficiency of this part is mainly affected by the graphics card GPU Performance impact, so compare the time consumption of different GPU models. The details are shown in Table 1.
From this we can see that the GeForce8800GT graphics card can basically complete the display function, and the GeForce GTX470 can display 35 to 40 pictures per second, which is a completely smooth video experience for users.
3 Conclusion
This year, the demand for computer operation demonstrations on large screens is increasing, and the development of high-performance graphics cards has prompted GPU computing to gradually become an important solution for large-scale parallel computing. The system uses CUDA technology to realize the video splicing system. At present, the system uses two dual-head graphics cards to form a display function module to achieve up to four screens. If you need more screens to display, you can consider using the network CS structure to expand. Because the time-consuming image processing part is completely calculated by the GPU, the more screens need to be calculated, and the number of system graphics cards will also increase, so the system will not degrade with the increase of the spliced ​​screen. Since the system uses parallel technology in general, it is easy to add the popular human-computer interaction module and photorealistic rendering module to the system in the future, making it a high-performance multimedia display system and giving users a full range of real experience .
Glass Display,High Transparency Led Glass Digital Screen,Glass Window Led Advertising Display,High Level Outdoor Glass Led Display
Kindwin Technology (H.K.) Limited , https://www.ktlleds.com