A real-time three-dimensional reconstruction system of moving objects based on fringe projection

Abstract

A kind of real-time three-dimensional reconstruction system of moving object based on stripe projection, comprising: processing system (PS end) for communicating with camera and projector, the acquisition and processing of image data are managed;Programmable logic (PL end) cooperates with processing system (PS end), is specially used for the acceleration of three-dimensional reconstruction, and the programmable logic (PL end) includes three-dimensional reconstruction algorithm hardware acceleration unit, and the real-time three-dimensional reconstruction algorithm of moving object in dynamic scene is executed by hardware acceleration;Wherein, the programmable logic (PL end) adopts stream computing architecture, and the received image data is handled by no cache stream design, and image data is analyzed instantly when each frame image arrives, and high-quality moving object real-time three-dimensional reconstruction result is continuously produced.The present application can not only keep high precision when processing three-dimensional reconstruction in dynamic scene, but also can realize fast reconstruction speed, effectively solve the challenge of reconstruction precision and speed in prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to real-time three-dimensional reconstruction technology, in particular to a moving object real-time three-dimensional reconstruction system based on stripe projection. BACKGROUND

[0002] In the field of computer vision and image processing, real-time three-dimensional reconstruction technology is an important research direction, and is widely used in virtual reality, augmented reality, robot navigation and industrial quality detection and other fields. Stripe projection is a commonly used three-dimensional reconstruction method, which uses a projection device to project a known stripe pattern onto the surface of an object, and then obtains the deformed stripe image through a camera device, and then calculates the three-dimensional information of the object through image analysis technology.

[0003] The current three-dimensional reconstruction system aims to achieve efficient reconstruction speed and superior reconstruction accuracy. Most systems use chips such as FPGA for data processing to significantly speed up the calculation. In these systems, image acquisition is completed by a high-speed camera, which can capture rapidly changing scenes at a high frame rate. The projection device is responsible for accurately projecting complex stripe patterns onto the surface of the object. The data is quickly processed by the FPGA to calculate high-precision three-dimensional reconstruction data.

[0004] However, the current three-dimensional reconstruction system based on stripe projection still faces some challenges, mainly focusing on the following two aspects, reconstruction accuracy and reconstruction speed. First, most current three-dimensional reconstruction systems are mainly optimized for static objects. When the system is applied to moving objects, the reconstruction accuracy is significantly reduced. This is because the surface features of moving objects change in a short period of time, and the deformation speed of the stripe pattern exceeds the capture capability of the camera, resulting in inaccurate three-dimensional reconstruction results. Second, existing three-dimensional reconstruction systems generally use a design that caches pictures first and then performs three-dimensional reconstruction. This strategy can ensure relatively comprehensive data collection, but also has significant problems: on the one hand, this method occupies a large amount of storage resources, especially under high frame rate and high resolution settings, the data size is more massive; on the other hand, the data transfer and processing link often becomes the bottleneck of system performance, slowing down the overall three-dimensional reconstruction speed. Therefore, in the application scenarios of fast feedback and real-time processing, this design greatly limits the effectiveness and application range of the system.

[0005] It should be noted that the information disclosed in the above background section is only for understanding the background of the present application, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The main purpose of the present application is to overcome the defects existing in the background art, and to provide a moving object real-time three-dimensional reconstruction system based on stripe projection.

[0007] To achieve the above object, the application adopts the following technical solutions:

[0008] A real-time three-dimensional reconstruction system of moving objects based on fringe projection, comprising:

[0009] A processing system (PS end) for communicating with a camera and a projector, managing the collection and processing of image data;

[0010] A programmable logic (PL end) cooperating with the processing system (PS end) and specially used for the acceleration of three-dimensional reconstruction, the programmable logic (PL end) comprising a three-dimensional reconstruction algorithm hardware acceleration unit, which executes the real-time three-dimensional reconstruction algorithm of moving objects in a dynamic scene in a hardware accelerated manner;

[0011] The programmable logic (PL end) adopts a streaming computing architecture and processes the received image data through a cacheless streaming design, analyzes the image data as soon as each frame of image arrives, and continuously generates high-quality real-time three-dimensional reconstruction results of moving objects.

[0012] Further, the three-dimensional reconstruction algorithm hardware acceleration unit comprises a fixed-point number calculation unit for replacing floating-point number calculation with fixed-point number calculation, optimizing the occupation of calculation resources and improving the calculation accuracy.

[0013] Further, the three-dimensional reconstruction algorithm hardware acceleration unit further comprises a data bit width allocation unit for accurately calculating and allocating the data bit width required by each numerical value, representing the numerical value with the smallest bit width while meeting the algorithm accuracy requirements.

[0014] Further, the cacheless streaming design of the streaming computing architecture comprises:

[0015] An image data rearrangement unit for rearranging the image data and concentrating the pixels at the same position in a group, optimizing the streaming processing efficiency.

[0016] Further, the image data rearrangement unit comprises:

[0017] A direct transmission module for directly transmitting the pixels at the same position of each image as a group from the processing system (PS end) to the programmable logic (PL end) without storing the image data in the BRAM cache;

[0018] A Gray code pattern projection unit for projecting Gray code patterns onto the surface of a target object, each pattern corresponding to a unique Gray code value, thereby roughly inferring the approximate depth interval of the pixel point;

[0019] The phase shift code pattern projection unit is used for projecting a series of phase shift code patterns with gradually changed phase difference to the surface of the target object after determining the approximate depth interval of the pixel points, and accurately calculating the depth information according to the change rule.

[0020] Further, the cache-free streaming design of the streaming computing architecture comprises:

[0021] The single-row pixel window processing unit is used for converting the data stream from the multi-row window pixel form into the single-row pixel window form, eliminating the data dependence between rows, and avoiding the bottleneck caused by waiting for the previous row data during processing.

[0022] Further, the single-row pixel window processing unit comprises:

[0023] The jump point detection module is used for detecting the jump point in the three-dimensional reconstruction calculation process, checking the gray value in the window centered on the three-dimensional reconstruction point by using the 1*n window smoothing method, and judging whether the point is greater than the average value of the remaining points to identify the jump point.

[0024] The jump point correction module is used for correcting most of the jump points by performing horizontal direction jump checking on the detected jump points at the PL end, and correcting the vertical direction jump points in the processing system (PS end).

[0025] Further, the streaming computing architecture adopts a multi-stage pipeline on the basis of the cache-free streaming design, subdivides the reconstruction process into a plurality of processing stages, and independently processes the specified tasks in each processing stage, so as to optimize the use of data flow and computing resources.

[0026] Further, each processing stage of the multi-stage pipeline has a double cache mechanism, which allows data reading in one cache area while data writing and processing in the other cache area, so as to reduce processing delay and improve the throughput of the system.

[0027] The present application has the following beneficial effects:

[0028] The present application aims at the problem of low three-dimensional reconstruction accuracy in a motion scene, and proposes a high-efficiency solution based on hardware implementation, which implements a motion scene three-dimensional reconstruction algorithm on hardware and maintains the high-precision advantage of the algorithm. For the problem of three-dimensional reconstruction speed, the present application proposes a cache-free streaming calculation architecture, which can effectively reduce data transfer. When each frame of image arrives, the PL end can immediately analyze it, thereby continuously generating high-quality three-dimensional reconstruction results, and avoiding the use of a large amount of on-chip storage resources. The present application has the advantages of higher precision and faster speed compared with the traditional background technology. Since the motion scene three-dimensional reconstruction algorithm is implemented on hardware, and the data bit width is designed in detail to ensure that the three-dimensional reconstruction precision on hardware is equivalent to the software three-dimensional reconstruction precision. By using the cache-free streaming calculation architecture, each frame of image data is processed in real time, the data transfer time is reduced, the three-dimensional reconstruction speed is improved, and the resource consumption is reduced.

[0029] Further, in the hardware implementation process of the three-dimensional reconstruction algorithm, fixed-point number calculation is used to replace floating-point number calculation in the software algorithm, the data bit width required for accurate calculation of each number is calculated, and a reasonable storage bit width is divided, so that each number value is represented by the smallest bit width, while still meeting the accuracy requirements of the algorithm.

[0030] Further, the multi-stage pipeline design allows each processing stage to work in parallel, thereby improving the overall throughput of the system. The double cache design makes the processing of each stage almost unaffected by the processing time of the previous and subsequent stages, greatly reducing the processing delay.

[0031] Other beneficial effects of the embodiments of the present application will be further described below. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 It is a structural diagram of the motion object real-time three-dimensional reconstruction system based on stripe projection in the embodiments of the present application.

[0033] Figure 2 It is a schematic diagram of the image data rearrangement unit in the embodiments of the present application.

[0034] Figure 3 It is a schematic diagram of the single-row pixel window processing unit in the embodiments of the present application.

[0035] Figure 4 It is a schematic diagram of the Gray code pattern projection in the embodiments of the present application.

[0036] Figure 5 It is a schematic diagram of the phase shift code pattern projection in the embodiments of the present application. DETAILED DESCRIPTION

[0037] The embodiments of the present application will be described in detail below. It should be emphasized that the following description is only exemplary and is not intended to limit the scope of the present application and its applications.

[0038] In addition, the terms "first", "second", "third", etc. are used only for descriptive purposes and are not to be construed as indicating or implying relative importance or an indicated number of technical features. Thus, features defined with "first", "second", etc. can explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0039] The principle of stripe projection three-dimensional reconstruction is mainly based on projecting a stripe pattern onto the surface of an object and obtaining the deformation effect of the object surface through a camera. By analyzing the changes of the projected stripes, the three-dimensional shape of the object can be recovered. However, in a moving scene, due to the rapid movement of the object and the change of the environment, the accuracy of three-dimensional reconstruction is often reduced, and the traditional algorithm repeatedly moves data when processing data, resulting in slow reconstruction speed. In order to overcome these challenges, the present application aims to solve the problem of reduced three-dimensional reconstruction accuracy in a moving scene, and proposes to implement a three-dimensional reconstruction algorithm for a moving scene on hardware while maintaining the high accuracy advantage of the algorithm. In addition, for the problem of three-dimensional reconstruction speed, the present application proposes a cache-free streaming computing architecture design, which can effectively reduce data movement. When each frame of image arrives, the PL end can immediately analyze it, thereby continuously generating high-quality three-dimensional reconstruction results, and avoiding the use of a large amount of on-chip storage resources.

[0040] Referring to Figure 1 , the embodiment of the present application provides a stripe projection-based real-time three-dimensional reconstruction system for moving objects, which comprises: a processing system (PS end) for communicating with a camera and a projector, managing the collection and processing of image data; a programmable logic (PL end) cooperating with the processing system (PS end) and specially used for the acceleration of three-dimensional reconstruction, the programmable logic (PL end) comprising a three-dimensional reconstruction algorithm hardware acceleration unit, which executes a real-time three-dimensional reconstruction algorithm for moving objects in a dynamic scene through hardware acceleration; wherein the programmable logic (PL end) adopts a streaming computing architecture and processes the received image data through a cache-free streaming design, analyzes the image data immediately when each frame of image arrives, and continuously generates high-quality real-time three-dimensional reconstruction results for moving objects.

[0041] In a preferred embodiment, the three-dimensional reconstruction algorithm hardware acceleration unit can include a fixed-point number calculation unit for replacing floating-point number calculation with fixed-point number calculation to optimize the occupation of calculation resources and improve the calculation accuracy. The three-dimensional reconstruction algorithm hardware acceleration unit can also include a data bit width allocation unit for accurately calculating and allocating the data bit width required for each numerical value to represent the numerical value with the smallest bit width while meeting the algorithm accuracy requirements.

[0042] In a preferred embodiment, the cache-free streaming design of the streaming computing architecture can include an image data rearrangement unit for rearranging image data to concentrate pixels of the same position in a group to optimize the efficiency of streaming processing. Particularly preferably, the image data rearrangement unit includes a direct transmission module for directly transmitting, at the processing system (PS side), pixels of the same position of each image as a group to the programmable logic (PL side) without storing the image data in the BRAM cache; a Gray code pattern projection unit for projecting Gray code patterns to the surface of the target object, each pattern corresponding to a unique Gray code value, thereby roughly inferring the approximate depth interval of the pixel point; and a phase shift code pattern projection unit for projecting, after determining the approximate depth interval of the pixel point, a series of phase shift code patterns with gradually changing phase differences to the surface of the target object and accurately calculating the depth information thereof according to the change law.

[0043] In a preferred embodiment, the cache-free streaming design of the streaming computing architecture can further include a single-row pixel window processing unit for converting the data stream from a multi-row window pixel form to a single-row pixel window form, eliminating the inter-row data dependency and avoiding bottlenecks due to waiting for the previous row of data during processing. Particularly preferably, the single-row pixel window processing unit includes a jump point detection module for detecting jump points in the three-dimensional reconstruction calculation process, checking the gray values in the window centered on the three-dimensional reconstruction point using a 1*n window smoothing method, and determining whether the point is greater than the average value of the remaining points to identify the jump point; and a jump point correction module for correcting most of the jump points by performing only horizontal direction jump checking at the PL side for the detected jump points, and correcting the vertical direction jump points in the processing system (PS side).

[0044] In a preferred embodiment, the streaming computing architecture adopts a multi-stage pipeline based on the cache-free streaming design, subdivides the reconstruction process into multiple processing stages, and processes the specified tasks independently in each processing stage to optimize the use of data flow and computing resources. Further, each processing stage of the multi-stage pipeline has a double cache mechanism, allowing data reading in one cache area while data writing and processing in the other cache area, to reduce processing delay and improve the throughput of the system.

[0045] The application provides a high-efficiency stripe projection-based real-time three-dimensional reconstruction system for moving objects, which significantly improves the speed and accuracy of three-dimensional reconstruction through the design of hardware acceleration and streaming computing architecture. The programmable logic (PL end) is introduced to cooperate with the processing system (PS end) to execute the real-time three-dimensional reconstruction algorithm for moving objects in a dynamic scene through hardware acceleration. The programmable logic (PL end) uses a cache-free streaming computing architecture to instantly analyze the image data of each frame and continuously generate high-quality three-dimensional reconstruction results, thereby avoiding the delay and resource consumption caused by caching a large amount of image data in the traditional design.

[0046] Further, the fixed-point number calculation is used to replace the traditional floating-point number calculation, which not only optimizes the occupation of computing resources but also improves the calculation accuracy. Through accurate data bit width allocation, each value is represented with the minimum bit width, which meets the algorithm accuracy requirement and reduces the storage resource requirement.

[0047] Further, the rearrangement of image data and the single-row pixel window processing further optimize the data flow, eliminate the inter-row data dependency, reduce the processing bottleneck, and improve the processing efficiency.

[0048] In addition, the multi-stage pipeline design and double buffer mechanism introduced allow the parallel work of various processing stages, greatly reduce the processing delay, and improve the overall throughput of the system.

[0049] The three-dimensional reconstruction system architecture of one specific embodiment is shown in Figure 1 The system is divided into two main parts: the processing system (PS end) and the programmable logic (PL end). In the PS end, the controller is responsible for communication with the camera and the projector, and manages the image data acquisition and processing. The camera captures stripe projection images in a moving scene in real time, and continuously transmits these images to the PL end for further processing. The PL end focuses on the acceleration of three-dimensional reconstruction, uses a dynamic scene three-dimensional reconstruction algorithm for hardware acceleration, and processes the received image data through a cache-free streaming design.

[0050] In the hardware implementation process of the three-dimensional reconstruction algorithm, balancing high accuracy and low resource occupation is a key challenge. Therefore, fixed-point number calculation is used to replace floating-point number calculation in the software algorithm, the data bit width required for each number is accurately calculated, and a reasonable storage bit width is allocated to ensure that each value is represented with the minimum bit width while still meeting the accuracy requirements of the algorithm.

[0051] In the streaming design of the three-dimensional reconstruction algorithm, the image data is rearranged and the data dependency is reduced from the algorithm level to achieve more efficient streaming processing. The image rearrangement design concentrates the pixels at the same position in a group, such as Figure 3As shown in the figure, thereby avoiding storing large blocks of image data in the cache of the PL end. Gray code is a special binary encoding method, as shown in the figure Figure 4 Gray code patterns are projected onto the surface of the target object, each pattern corresponds to a unique Gray code value, thereby roughly inferring the approximate depth interval of the pixel point, after determining the approximate depth interval of the pixel point, a series of phase difference gradually changing phase shift code patterns are projected onto the surface of the target object, as shown in the figure Figure 5 The depth information is accurately calculated according to the change rule. The image rearrangement method used in the prior art is to divide a large image into small blocks, as shown in the figure Figure 2 As shown in the figure, a plurality of rows of each image are extracted as a group and cached in the BRAM of the PL end, and each time the pixels at the same position of each image are read from the BRAM as a group for subsequent three-dimensional reconstruction calculation, the image rearrangement method used in the present application is as shown in the figure Figure 3 The pixels at the same position of each image are transmitted to the PL end as a group in the processing system (PS end), and subsequent three-dimensional reconstruction calculation is directly performed, without the need for BRAM caching.

[0052] In addition, the data stream originally processed in the form of multiple row window pixels is changed to the form of single row pixel window. This optimization effectively eliminates the data dependency between rows, ensures that there is no bottleneck during processing due to waiting for the previous row of data, thereby avoiding the problem of multi-row caching, as shown in the figure Figure 3 In the three-dimensional reconstruction calculation process, there may be some reconstructed three-dimensional points that deviate from the original surface, referred to as jump points. In order to reduce the occurrence of jump points, a n*n window smoothing method is often used, that is, whether the point is greater than the average value of the remaining points in the window centered on the three-dimensional reconstruction point is detected, and if it is greater, it is a jump point. Therefore, there is a data dependency between rows. In view of this, the n*n window smoothing of the PL end is optimized to 1*n window smoothing. Since the bright and dark transitions of the stripes are easy to calculate jump points, both Gray code and phase shift code are vertical stripes. Generally, the gray value is continuous in the vertical direction and discontinuous in the horizontal direction. The PL only needs to check the horizontal direction to correct most of the jump points, and the vertical direction jump points are corrected in the PS end.

[0053] The present application proposes a high-efficiency solution combining hardware implementation to solve the problems of low three-dimensional reconstruction accuracy and slow speed in motion scenes. Through the three-dimensional reconstruction system constructed on the Zynq development board, the processing system (PS end) and the programmable logic (PL end) are used to acquire and process the stripe projection images in motion scenes in real time. In the design, fixed-point number calculation and accurate data bit width allocation are used to achieve the balance between high precision and low resource occupation, and a cache-free streaming calculation architecture is introduced to reduce data transfer and improve processing speed. In addition, the efficiency of streaming processing is optimized through the rearrangement of image data and the elimination of row data dependency.

[0054] To further improve efficiency and accuracy in processing three-dimensional reconstruction in dynamic scenes, on the basis of the streaming design, a multi-stage pipeline design scheme is proposed. This scheme subdivides the three-dimensional reconstruction process into multiple stages, and each stage handles specific tasks independently to optimize data flow and computing resource usage. On the basis of the three-dimensional reconstruction process being broken down into multiple independent processing stages, further, for each processing stage, a double buffer mechanism is adopted. This means that there are two buffer areas for the intermediate data of each stage of processing, so that while one buffer area is reading data, the other buffer area is writing and processing data, ensuring that there is no waiting or delay during processing. The advantage of this scheme is that the multi-stage pipeline design allows the various processing stages to work in parallel, thereby improving the overall throughput of the system. The double buffer design makes the processing of each stage almost unaffected by the processing time of the previous and subsequent stages, greatly reducing the processing delay.

[0055] In summary, the present application proposes a hardware-based efficient solution for the stripe projection three-dimensional reconstruction technology, which has the advantages of higher precision and faster speed compared to traditional background technology. First, the motion scene three-dimensional reconstruction algorithm is implemented on hardware, and the data bit width is designed finely to ensure that the three-dimensional reconstruction precision on hardware is comparable to that of software. Second, a cache-free streaming computing architecture is adopted to process image data in real time, reducing data transfer time, improving three-dimensional reconstruction speed, and reducing resource usage.

[0056] The present application not only maintains high precision, but also achieves fast reconstruction speed when processing three-dimensional reconstruction in dynamic scenes, effectively solving the challenges of reconstruction precision and speed in the prior art, especially in the application scenarios of fast feedback and real-time processing, significantly improving the effectiveness and application range of the system.

[0057] The above further describes the present application in conjunction with specific / preferred embodiments, and cannot be deemed to limit the specific implementation of the present application to these descriptions. For those skilled in the art to which the present application belongs, without departing from the concept of the present application, they can make several substitutions or variations to the described embodiments, and these substitutions or variations shall be deemed to fall within the protection scope of the present application. In the description of the present application, the description of the terms "an embodiment", "some embodiments", "a preferred embodiment", "an example", "a specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are contained in at least one embodiment or example of the present application. In the present application, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In the case of no mutual contradiction, those skilled in the art can combine and combine the different embodiments or examples described in the present application and the features of the different embodiments or examples. Although the embodiments of the present application and their advantages have been described in detail, it should be understood that various changes, substitutions and modifications can be made herein without departing from the scope of protection of the patent application.

Claims

1. A stripe projection based real-time three-dimensional reconstruction system of moving objects, characterized in that, The application relates to a three-dimensional reconstruction system, comprising: a processing system PS end for communicating with a camera and a projector, managing the collection and processing of image data; a programmable logic PL end cooperating with the processing system PS end, dedicated to the acceleration of three-dimensional reconstruction, the programmable logic PL end comprising a three-dimensional reconstruction algorithm hardware acceleration unit for executing a real-time three-dimensional reconstruction algorithm of a moving object in a dynamic scene through hardware acceleration; the three-dimensional reconstruction algorithm hardware acceleration unit comprising: a fixed-point number calculation unit for replacing floating-point number calculation with fixed-point number calculation; and a data bit width allocation unit for calculating and allocating the data bit width required by each value, so that the value is represented by the minimum bit width while meeting the algorithm accuracy requirement; wherein the programmable logic PL end adopts a streaming computation architecture, processes the received image data through a cache-free streaming design, analyzes the image data as soon as each frame of image arrives, and continuously generates high-quality real-time three-dimensional reconstruction results of the moving object; the cache-free streaming design of the streaming computation architecture comprises: an image data rearrangement unit for rearranging the image data, concentrating the pixels at the same position in a group, and optimizing the streaming processing efficiency; the image data rearrangement unit comprises: a direct transmission module for directly transmitting the pixels at the same position of each image as a group from the processing system PS end to the programmable logic PL end without storing the image data in a BRAM cache; and a Gray code pattern projection unit for projecting a Gray code pattern onto the surface of a target object, each pattern corresponding to a unique Gray code value, so as to infer the depth interval of the pixel point; and a phase shift code pattern projection unit for projecting a phase shift code pattern with gradually changing phase difference onto the surface of the target object after the approximate depth interval of the pixel point is determined, and calculating the depth information according to the change rule.

2. The stripe projection based real-time three-dimensional reconstruction system of moving objects of claim 1, wherein, The cache-free streaming design of the streaming computation architecture comprises: a single-row pixel window processing unit for converting the data stream from a multi-row window pixel form into a single-row pixel window form, and eliminating the inter-row data dependence.

3. The stripe projection based real-time three-dimensional reconstruction system of moving objects of claim 2, wherein, The single-row pixel window processing unit comprises: a jump point detection module for detecting jump points in the three-dimensional reconstruction calculation process, checking the gray values in the window centered on the three-dimensional reconstruction point in a 1*n window smoothing manner, judging whether the point is greater than the average value of the remaining points, and identifying the jump points; a jump point correction module for correcting most of the jump points by performing horizontal direction jump checking on the detected jump points in the PL end, and correcting the vertical direction jump points in the processing system PS end.

4. The stripe projection based real-time 3D reconstruction system of moving objects according to any of claims 1 to 3, characterized in that, The streaming computation architecture adopts a multi-stage pipeline based on the cache-free streaming design, subdivides the reconstruction process into multiple processing stages, and independently processes the specified tasks in each processing stage, so as to optimize the use of data flow and computing resources.

5. The fringe projection based real-time 3D reconstruction system of moving objects of claim 4, wherein, Each processing stage of the multi-stage pipeline has a double cache mechanism, allowing data reading in one cache area while data writing and processing in the other cache area, so as to reduce processing delay and improve the throughput of the system.

Images