Unlock instant, AI-driven research and patent intelligence for your innovation.

Guided-filtering optimization speed-up method based on CUDA

An optimization method and guided filtering technology, applied in image data processing, instruments, calculations, etc., can solve the problem that it is difficult to meet the requirements of algorithm accuracy and real-time processing, guided filtering algorithm consumes computing time and space, and is difficult to achieve algorithm accuracy Algorithm execution efficiency and other issues, to achieve the effect of low hardware requirements, great innovation significance, and novel ideas

Active Publication Date: 2015-09-09
天津渤化安创科技有限公司
View PDF3 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This algorithm is simple and effective, but it needs to calculate complex matrices and solve large linear equations, resulting in a large amount of computing time and space for the guided filtering algorithm, which cannot meet the needs of practical applications
[0004] All in all, the guided image filtering algorithm has a large amount of calculation, and it is difficult to improve the execution efficiency of the algorithm while ensuring the accuracy of the algorithm
Therefore, the traditional CPU-based architecture is difficult to meet people's requirements for algorithm accuracy and real-time processing, and only the graphics processor (GPU) is used to meet the needs of practical applications.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Guided-filtering optimization speed-up method based on CUDA
  • Guided-filtering optimization speed-up method based on CUDA
  • Guided-filtering optimization speed-up method based on CUDA

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] A CUDA-based guided filtering acceleration optimization method, see figure 1 , the guided filtering acceleration optimization method includes the following steps:

[0035] 101: Read the input image p and the guide image I from the host-side memory into the global memory, and obtain the input image p, the guide image I, the image I*P, and the image I*I in the neighborhood window respectively by constructing the first kernel function. image neighborhood mean;

[0036] 102: Construct the second kernel function to obtain the covariance of the image (I, p) in turn, guide the variance of the image I, and then obtain the key filtering parameters a and b;

[0037] 103: Call the first kernel function to obtain the neighborhood mean value mean_a of parameter a and the neighborhood mean value of parameter b

[0038] mean_b, and then obtain the final filtering result, save the result to the corresponding global memory, and transfer it to the host-side memory.

[0039] Wherein, b...

Embodiment 2

[0049] The guided filtering algorithm is implemented based on a local linear model. In the local linear model, let the input image be p, the guiding image be I, and the filtered output image be q. The local linear model assumes that the neighborhood window ω with the center pixel k k There is a linear relationship as follows:

[0050] q i = a k I i + b k , ∀ i ∈ ω k

[0051] (1)

[0052] Among them, ω k is a square window with side length r, a k and b k is the neighborhood window ω k The linear coefficient in, I i For the guide image in the neighborhood window ω k Pixel value in q i is the neighborhood window ω k Filtered output in . Coefficient a k and b k It can be determined by finding the minimum difference between the input...

Embodiment 3

[0092] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further described in detail below in conjunction with specific examples.

[0093] The example of the present invention adopts windows 7 operating system, and CPU is Intel Core i5-3470, and main frequency is 3.2GHz, and system memory is 4GB; Contains 192 CUDA cores, onboard global memory is 2048Mbytes, memory bandwidth is 192 bits, and supports CUDA Compute Capability 3.0. At the same time, the present invention utilizes the Visual Profile that comes with the CUDA Toolkit to analyze various data to realize the quantitative analysis of the program performance.

[0094] In order to verify the effectiveness of this method, the example of the present invention carries out CUDA parallel optimization to the guided filtering algorithm in four application fields such as image smoothing, image feathering, image enhancement and f...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a guided-filtering optimization speed-up method based on CUDA, and the method comprises the following steps: enabling an input image p and a guide image I to be read into a global storage unit from a memory of a host end; respectively obtaining neighborhood mean values of the input image p, the guide image I, an image I*P and an image I*I at neighborhood windows through the construction of a first core function; constructing a second core function, and sequentially obtaining the covariance of images (I, p) and the covariance of the image I, thereby obtaining the key parameters a and b of filtering; performing the call of the first core function to obtain a neighborhood mean value mean_a of the parameter a and the neighborhood mean value mean_b of the parameter b, thereby obtaining the final filtering result q; enabling the result q to be stored in the corresponding global storage unit, and outputting the result q to the memory of the host end. The method employs the advantages in the floating point calculation and parallel computing capability of a GPU, guarantees the filtering effect of an image, and effectively improves the execution efficiency of a guided filtering algorithm, and quickly achieves the guided filtering algorithm.

Description

technical field [0001] The invention relates to the fields of computer application technology and image processing, in particular to a CUDA (Unified Computing Device Architecture)-based guided filtering acceleration optimization method. Background technique [0002] Image filtering is an important means of image processing, which has great significance and research value. Due to the imperfection of imaging system, transmission medium and recording equipment, digital images are often polluted by various noises in the process of formation, transmission and recording. Image filtering, that is, to suppress the noise of the target image under the condition of retaining the image details as much as possible, is an indispensable operation in image preprocessing, and its processing effect will directly affect the effectiveness of subsequent image processing and analysis. and reliability. [0003] Image filtering methods can be divided into two types: one is linear shift-invariant ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T5/00
Inventor 何凯王新磊王晓文葛云峰
Owner 天津渤化安创科技有限公司