Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures
Application of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures is described. The description of realization of algorithm of search of images' defects by the means of technology CUDA (Compute Unified Device Architecture - the unified hardware-soft...
Saved in:
| Date: | 2013 |
|---|---|
| Main Authors: | , , , , |
| Format: | Article |
| Language: | English |
| Published: |
Національний науковий центр «Харківський фізико-технічний інститут» НАН України
2013
|
| Series: | Вопросы атомной науки и техники |
| Subjects: | |
| Online Access: | https://nasplib.isofts.kiev.ua/handle/123456789/111837 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Journal Title: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Cite this: | Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures / V.A. Dudnik, V.I. Kudryavtsev, T.M. Sereda, S.A. Us, M.V. Shestakov // Вопросы атомной науки и техники. — 2013. — № 3. — С. 282-284. — Бібліогр.: 6 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
nasplib_isofts_kiev_ua-123456789-111837 |
|---|---|
| record_format |
dspace |
| spelling |
nasplib_isofts_kiev_ua-123456789-1118372025-02-09T20:52:08Z Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures Використання засобiв GPGPU для розробки програм пошуку дефектiв монохромних пiвтонових зображень Использование средств GPGPU для разработки программ поиска дефектов монохромных полутоновых изображений Dudnik, V.A. Kudryavtsev, V.I. Sereda, T.M. Us, S.A. Shestakov, M.V. Вычислительные и модельные системы Application of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures is described. The description of realization of algorithm of search of images' defects by the means of technology CUDA (Compute Unified Device Architecture - the unified hardware-software decision for parallel calculations on GPU) companies NVIDIA is resulted. It is done the comparison of the temporary characteristics of performance of images' updating without application GPU and with use of opportunities of graphic processor GeForce 8800. Описано застосування засобiв GPGPU для розробки програм пошуку дефектiв монохромних пiвтонових зображень. Приведено опис реалiзацiї алгоритму пошуку дефектiв зображень засобами технологiї CUDA (Compute Unified Device Architecture – унiфiкованого програмно-апаратного рiшення для паралельних обчислень на GPU) компанiї NVIDIA. Прведено порiвняння тимчасових характеристик виконання коректування зображень без застосування GPU i з використанням можливостей графiчного процесора GeForce 8800. Описано применение средств GPGPU для разработки программ поиска дефектов монохромных полутоновых изображений. Приведено описание реализации алгоритма поиска дефектов изображений средствами технологии CUDA (Compute Unified Device Architecture – унифицированного программно- аппаратного решения для параллельных вычислений на GPU) компании NVIDIA. Проведено сравнение временных характеристик выполнения корректировки изображений без применения GPU и с использованием возможностей графического процессора GeForce 8800. 2013 Article Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures / V.A. Dudnik, V.I. Kudryavtsev, T.M. Sereda, S.A. Us, M.V. Shestakov // Вопросы атомной науки и техники. — 2013. — № 3. — С. 282-284. — Бібліогр.: 6 назв. — англ. 1562-6016 PACS: 89.80.+h, 89.70.+c, 01.10.Hx https://nasplib.isofts.kiev.ua/handle/123456789/111837 en Вопросы атомной науки и техники application/pdf Національний науковий центр «Харківський фізико-технічний інститут» НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| topic |
Вычислительные и модельные системы Вычислительные и модельные системы |
| spellingShingle |
Вычислительные и модельные системы Вычислительные и модельные системы Dudnik, V.A. Kudryavtsev, V.I. Sereda, T.M. Us, S.A. Shestakov, M.V. Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures Вопросы атомной науки и техники |
| description |
Application of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures is described. The description of realization of algorithm of search of images' defects by the means of technology CUDA (Compute Unified Device Architecture - the unified hardware-software decision for parallel calculations on GPU) companies NVIDIA is resulted. It is done the comparison of the temporary characteristics of performance of images' updating without application GPU and with use of opportunities of graphic processor GeForce 8800. |
| format |
Article |
| author |
Dudnik, V.A. Kudryavtsev, V.I. Sereda, T.M. Us, S.A. Shestakov, M.V. |
| author_facet |
Dudnik, V.A. Kudryavtsev, V.I. Sereda, T.M. Us, S.A. Shestakov, M.V. |
| author_sort |
Dudnik, V.A. |
| title |
Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures |
| title_short |
Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures |
| title_full |
Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures |
| title_fullStr |
Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures |
| title_full_unstemmed |
Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures |
| title_sort |
use of a gpgpu means for the development of search programs of defects of monochrome half-tone pictures |
| publisher |
Національний науковий центр «Харківський фізико-технічний інститут» НАН України |
| publishDate |
2013 |
| topic_facet |
Вычислительные и модельные системы |
| url |
https://nasplib.isofts.kiev.ua/handle/123456789/111837 |
| citation_txt |
Use of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures / V.A. Dudnik, V.I. Kudryavtsev, T.M. Sereda, S.A. Us, M.V. Shestakov // Вопросы атомной науки и техники. — 2013. — № 3. — С. 282-284. — Бібліогр.: 6 назв. — англ. |
| series |
Вопросы атомной науки и техники |
| work_keys_str_mv |
AT dudnikva useofagpgpumeansforthedevelopmentofsearchprogramsofdefectsofmonochromehalftonepictures AT kudryavtsevvi useofagpgpumeansforthedevelopmentofsearchprogramsofdefectsofmonochromehalftonepictures AT seredatm useofagpgpumeansforthedevelopmentofsearchprogramsofdefectsofmonochromehalftonepictures AT ussa useofagpgpumeansforthedevelopmentofsearchprogramsofdefectsofmonochromehalftonepictures AT shestakovmv useofagpgpumeansforthedevelopmentofsearchprogramsofdefectsofmonochromehalftonepictures AT dudnikva vikoristannâzasobivgpgpudlârozrobkiprogrampošukudefektivmonohromnihpivtonovihzobraženʹ AT kudryavtsevvi vikoristannâzasobivgpgpudlârozrobkiprogrampošukudefektivmonohromnihpivtonovihzobraženʹ AT seredatm vikoristannâzasobivgpgpudlârozrobkiprogrampošukudefektivmonohromnihpivtonovihzobraženʹ AT ussa vikoristannâzasobivgpgpudlârozrobkiprogrampošukudefektivmonohromnihpivtonovihzobraženʹ AT shestakovmv vikoristannâzasobivgpgpudlârozrobkiprogrampošukudefektivmonohromnihpivtonovihzobraženʹ AT dudnikva ispolʹzovaniesredstvgpgpudlârazrabotkiprogrammpoiskadefektovmonohromnyhpolutonovyhizobraženii AT kudryavtsevvi ispolʹzovaniesredstvgpgpudlârazrabotkiprogrammpoiskadefektovmonohromnyhpolutonovyhizobraženii AT seredatm ispolʹzovaniesredstvgpgpudlârazrabotkiprogrammpoiskadefektovmonohromnyhpolutonovyhizobraženii AT ussa ispolʹzovaniesredstvgpgpudlârazrabotkiprogrammpoiskadefektovmonohromnyhpolutonovyhizobraženii AT shestakovmv ispolʹzovaniesredstvgpgpudlârazrabotkiprogrammpoiskadefektovmonohromnyhpolutonovyhizobraženii |
| first_indexed |
2025-11-30T16:01:20Z |
| last_indexed |
2025-11-30T16:01:20Z |
| _version_ |
1850231736000053248 |
| fulltext |
COMPUTING AND MODELLING SYSTEMS
USE OF A GPGPU MEANS FOR THE DEVELOPMENT OF
SEARCH PROGRAMS OF DEFECTS OF MONOCHROME
HALF-TONE PICTURES
V.A.Dudnik,∗V.I.Kudryavtsev, T.M.Sereda, S.A.Us, M.V.Shestakov
National Science Center ”Kharkov Institute of Physics and Technology”, 61108, Kharkov, Ukraine
(Received January 16, 2012)
Application of a GPGPU means for the development of search programs of defects of monochrome half-tone pictures
is described. The description of realization of algorithm of search of images’ defects by the means of technology
CUDA (Compute Unified Device Architecture - the unified hardware-software decision for parallel calculations on
GPU) companies NVIDIA is resulted. It is done the comparison of the temporary characteristics of performance of
images’ updating without application GPU and with use of opportunities of graphic processor GeForce 8800.
PACS: 89.80.+h, 89.70.+c, 01.10.Hx
1. INTRODUCTION
Monochrome images are turned out as a result of
processing the data given by the sensor controls used
for medicine, flaw detection, special cartography and
etc. In this case the image is turned out on an output
of X-ray camera used for the control of temperature
parameters in the chamber of a high pressure press.
Prominent feature of such images is presence of spe-
cific defects which are very bright points.
Fig.1. A fragment of the original image with defects
Such defects are a result from hitting of the absent-
minded quantum of X-ray radiation directly on a ma-
trix of the sensor control. In this case because of
rather bigger size of the camera the voltage on the
anode of a used source of radiation reaches 250 kilo-
volt. It leads to the occurrence of rather big quantity
of quantum of the absent-minded radiation. Within
the analysis of similar images it is necessary the ap-
plication of the programs-filters removing the similar
defects. It slows down the formation of images essen-
tially (in several times). Acceleration of processing
is necessary for improvement of the analysis of such
images. Basic time when setting up of a window of
visibility is occupied by the filtration of images. It is
necessary for removing of such defects.
2. IMAGE FILTERING
Filters are based on the operation of convolution
[1,3,4]. An operation of calculation of a new value of
the chosen pixel is applied for the image of convolu-
tion. It considers the values of the pixels surrounding
it. For this case the matrix (a kernel of convolution)
by the size 3× 3 is used:
P(i-1,j-1) P(i,j-1) P(i+1,j-1)
P(i-1,j) P(i,j) P(i+1,j)
P(i-1,j+1) P(i,j+1) P(i+1,j+1)
If we apply convolution to each pixel of the image
the certain effect is turned out which depends on the
chosen kernel of convolution. Generally there are two
types of the filters on usage of pixels’ values: recur-
sive and non- recursive. In recursive filters the val-
ues calculated on the previous step are used for the
calculation of the subsequent pixels’ values. In non-
recursive filters for calculations it is always used the
original pixels’ values. It is necessary to note that it
is rather difficult to do paralleling of calculation for
the recursive filters. It is impossible for CUDA. That
is why the application of such filters is not considered
here.
3. MEDIAN FILTER
For the filtration of impulsive disturbances (in this
case they are bright points) median filters are usu-
ally used. It is one of a kind of digital filters which
∗Corresponding author. E-mail address: dudnik@mail.ru
282 ISSN 1562-6016. PROBLEMS OF ATOMIC SCIENCE AND TECHNOLOGY, 2013, N3(85).
Series: Nuclear Physics Investigations (60), p.282-284.
is widely used in digital processing of signals and im-
ages for the reduction of the noise level. It is based
on finding of a median which is an average element
of sequence. For this the value of readout inside of
the window of the filter are sorted in ascending order
(decrease). The value which is in the middle of the or-
dered list comes to the output of the filter. The win-
dow moves along filtered signal and calculations are
repeated. In this way very bright values of elements of
the image are eliminated and replaced by the values of
similar magnitudes of brightness of the next elements.
Fig.2. A fragment of the filtered image with
the removed defects
Use of the graphic accelerators as a fast calculators
allows to accelerate processing of the half-tone pic-
tures which have size more of 10 000 elements. Re-
alization of the programs of images processing in the
form of DLL is recommended. It allows to use them
for the various program’s platforms. The gap in pro-
ductivity between GPU and CPU is sharply reduced
(it is approximately proportional to a quantity of ker-
nels) with the usage of the multinuclear (2?6 kernels)
processors and increasing of calculations’ complexity
for paralleling processes.
4. PROGRAM REALIZATION OF
ALGORITHM
The algorithm has been realized by the means of
system engineering CUDA in the form of Windows
DLL. It provided a convenient reference to it from
the programs created by the means of various soft-
ware (for example: MS Visual Studio and Borland C
++ Builder). Moving of the data processing image
to the GPU device is done during the reference to
the program of filtration. Processing of the received
data by the means of median filter and returning of
the filtered image to the specified area of computer’s
memory is also carried out [2,5]. Below the fragment
of the code which is carrying out described above op-
eration is resulted:
#include <stdio.h>
#include <stdio.h>
#include <assert.h>
#include <assert.h>
#include <cuda.h>
#include <cuda.h>
int main(void)
int main(void)
{
// pointers to host memory
float *a_h, *b_h; float *a_h, *b_h; float *a_d, *b_d;
// pointers to device memory
float *a_d, *b_d; int i; int i;
// allocate arrays on host
a_h = (float *)malloc(sizeof(float)*N); a_h = (float
*)malloc(sizeof(float)*N); b_h = (float *)malloc(sizeof(float)*N);
b_h = (float *)malloc(sizeof(float)*N);
// allocate arrays on device
cudaMalloc((void **) &a_d, sizeof(float)*N); cudaMalloc((void **)
&a_d, sizeof(float)*N); cudaMalloc((void **) &b_d, sizeof(float)*N);
cudaMalloc((void **) &b_d, sizeof(float)*N);
// send data from host to device: a_h to a_d
cudaMemcpy(a_d, a_h, sizeof(float)*N, cudaMemcpyHostToDevice);
cudaMemcpy(a_d, a_h, sizeof(float)*N, cudaMemcpyHostToDevice);
// copy data within device: a_d to b_d
cudaMemcpy(b_d, a_d, sizeof(float)*N, cudaMemcpyDeviceToDevice);
cudaMemcpy(b_d, a_d, sizeof(float)*N, cudaMemcpyDeviceToDevice);
// Delete defects
MedianFilter(b_d);
// retrieve data from device: b_d to b_h
cudaMemcpy(b_h, b_d, sizeof(float)*N, cudaMemcpyDeviceToHost);
cudaMemcpy(b_h, b_d, sizeof(float)*N, cudaMemcpyDeviceToHost);
283
It is necessary to note that an overhead charge
for the transfer and returning of a file of processing
data, start of the process of processing occupy an es-
sential part of the general operating time of the pro-
gram. That is why an essential gain in time starts
to be shown within the size of processing files of 20
thousand of elements and more. In this case the
same algorithm has been realized with the help of
the base means of parallel processing which is avail-
able within Windows. The comparison analysis of
programs’ work speed has shown that the usage of
modern multinuclear processors with general purpose
allows making a break in speed of processing unim-
pressive (2/5 times in comparison with 10/20 for the
one-nuclear processor). On the other hand it has
been noted by us that the increase in processing speed
rather sharp decreases within increasing the quantity
of kernels more than seven. (From our point of view
it is due to increasing the overhead charge of Win-
dows system for the management of quantization of
time and switching of the processes, and as due to in-
terference of the data in the buffer of the processor).
5. CONCLUSIONS
Use of the graphic accelerators as a fast calculators
allows to accelerate processing of the half-tone pic-
tures which have size more of 10 000 elements. Re-
alization of the programs of images processing in the
form of DLL is recommended. It allows to use them
for the various program’s platforms. The gap in pro-
ductivity between GPU and CPU is sharply reduced
(it is approximately proportional to a quantity of ker-
nels) with the usage of the multinuclear (2?6 kernels)
processors and increasing of calculations’ complexity
for paralleling processes.
References
1. William K. Pratt. Digital Image Processing 3-rd
Edition. On 2011. Wiley-Interscience; 2 edition
(April 1991).
2. Aleksej Berillo. (NVIDIA CUDA) Non
graphic calculation on graphic processors.
http://www.ixbt.com/video3/cuda-1.shtml
3. Kenneth R. Castleman. Digital Image Process-
ing. Prentice Hall; 2nd edition (September 2,
1995).
4. Geoff Dougherty. Digital Image Processing for
Medical Applications. Cambridge University
Press; 1 edition (May 11, 2009).
5. Federico Dal Castello. Advanced System Tech-
nology. STMicroelectronics, Italy Douglas
Miles, The Portland Group: Parallel Ran-
dom Number Generation Using OpenMP,
OpenCL and PGI Accelerator Directives.
http://www.pgroup.com/lit/articles/insider/
v2n2a4.htm
6. Don Breazeal,Craig Toepfer: Tuning Ap-
plication Performance Using Hardware
Event Counters in the PGPROF Profiler
http://www.pgroup.com/lit/articles/insider/
v2n4a3.htm en-us/File/ larrabee-”manycore.pdf”
ИСПОЛЬЗОВАНИЕ СРЕДСТВ GPGPU ДЛЯ РАЗРАБОТКИ ПРОГРАММ ПОИСКА
ДЕФЕКТОВ МОНОХРОМНЫХ ПОЛУТОНОВЫХ ИЗОБРАЖЕНИЙ
В.А.Дудник, В.И.Кудрявцев, Т.М.Середа, С.А.Ус, М.В.Шестаков
Описано применение средств GPGPU для разработки программ поиска дефектов монохромных по-
лутоновых изображений. Приведено описание реализации алгоритма поиска дефектов изображений
средствами технологии CUDA (Compute Unified Device Architecture – унифицированного программно-
аппаратного решения для параллельных вычислений на GPU) компании NVIDIA. Проведено сравнение
временных характеристик выполнения корректировки изображений без применения GPU и с исполь-
зованием возможностей графического процессора GeForce 8800.
ВИКОРИСТАННЯ ЗАСОБIВ GPGPU ДЛЯ РОЗРОБКИ ПРОГРАМ ПОШУКУ
ДЕФЕКТIВ МОНОХРОМНИХ ПIВТОНОВИХ ЗОБРАЖЕНЬ
В.О.Дуднiк, В.I.Кудрявцев, Т.М.Середа, С.О.Ус, М.В.Шестаков
Описано застосування засобiв GPGPU для розробки програм пошуку дефектiв монохромних пiвто-
нових зображень. Приведено опис реалiзацiї алгоритму пошуку дефектiв зображень засобами техно-
логiї CUDA (Compute Unified Device Architecture – унiфiкованого програмно-апаратного рiшення для
паралельних обчислень на GPU) компанiї NVIDIA. Прведено порiвняння тимчасових характеристик
виконання коректування зображень без застосування GPU i з використанням можливостей графiчного
процесора GeForce 8800.
284
|