24

C++ Implementation of the Paper Median Filter in Constant Time

 4 years ago
source link: https://www.tuicool.com/articles/InE7zmu
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.

Median Filter In Constant Time

C++ implementation of the paper Median Filter In Constant Time . Successfully build on linux and android(tested on HUAWEI P30, HiSilicon KIRIN 980).

Build

Build on linux

  1. Install cmake 3.11

  2. Build the project by running the scripts/build_linux.sh script and remember to modify the following parameter. 

CMAKE_PATH=cmake
  1. After successfully building the project, the executable runUnitTests will be executed. If every thing goes right, you should see the following outputs on your terminal. 
[==========] Running 15 tests from 1 test suite.
[----------] Global test environment set-up.
[----------] 15 tests from netTest
[ RUN      ] netTest.medianFilterFloatRaduis1SpeedTest

#################Normal Float MedianFilfer Radius 1 Average Cost time: 11.6482ms

[       OK ] netTest.medianFilterFloatRaduis1SpeedTest (119 ms)
[ RUN      ] netTest.medianFilterUint16Raduis1SpeedTest

#################Normal Uint16 MedianFilfer Radius 1 Average Cost time: 10.5113ms

[       OK ] netTest.medianFilterUint16Raduis1SpeedTest (107 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis1SpeedTest

#################Constant time Uint16 MedianFilfer Radius 1 Average Cost time: 33.889ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis1SpeedTest (341 ms)
[ RUN      ] netTest.medianFilterFloatRaduis2SpeedTest

#################Normal Float MedianFilfer Radius 2 Average Cost time: 78.2531ms

[       OK ] netTest.medianFilterFloatRaduis2SpeedTest (784 ms)
[ RUN      ] netTest.medianFilterUint16Raduis2SpeedTest

#################Normal Uint16 MedianFilfer Radius 2 Average Cost time: 63.8422ms

[       OK ] netTest.medianFilterUint16Raduis2SpeedTest (641 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis2SpeedTest

#################Constant Time Uint16 MedianFilfer Radius 2 Average Cost time: 36.4003ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis2SpeedTest (365 ms)
[ RUN      ] netTest.medianFilterFloatRaduis3SpeedTest

#################Normal Float MedianFilfer Radius 3 Average Cost time: 265.695ms

[       OK ] netTest.medianFilterFloatRaduis3SpeedTest (2660 ms)
[ RUN      ] netTest.medianFilterUint16Raduis3SpeedTest

#################Normal Uint16 MedianFilfer Radius 3 Average Cost time: 213.539ms

[       OK ] netTest.medianFilterUint16Raduis3SpeedTest (2137 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis3SpeedTest

#################Constant time Uint16 MedianFilfer Radius 3 Average Cost time: 36.9943ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis3SpeedTest (371 ms)
[ RUN      ] netTest.medianFilterFloatRaduis4SpeedTest

#################Normal Float MedianFilfer Radius 4 Average Cost time: 683.465ms

[       OK ] netTest.medianFilterFloatRaduis4SpeedTest (6837 ms)
[ RUN      ] netTest.medianFilterUint16Raduis4SpeedTest

#################Normal Uint16 MedianFilfer Radius 4 Average Cost time: 545.849ms

[       OK ] netTest.medianFilterUint16Raduis4SpeedTest (5461 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis4SpeedTest

#################Constant time Uint16 MedianFilfer Radius 4 Average Cost time: 37.2009ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis4SpeedTest (373 ms)
[ RUN      ] netTest.medianFilterFloatRaduis5SpeedTest

#################Normal Float MedianFilfer Radius 5 Average Cost time: 1470.26ms

[       OK ] netTest.medianFilterFloatRaduis5SpeedTest (14705 ms)
[ RUN      ] netTest.medianFilterUint16Raduis5SpeedTest

#################Normal Uint16 MedianFilfer Radius 5 Average Cost time: 1179.28ms

[       OK ] netTest.medianFilterUint16Raduis5SpeedTest (11795 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis5SpeedTest

#################Constant time Uint16 MedianFilfer Radius 5 Average Cost time: 37.2977ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis5SpeedTest (375 ms)
[----------] 15 tests from netTest (47071 ms total)

[----------] Global test environment tear-down
[==========] 15 tests from 1 test suite ran. (47071 ms total)
[  PASSED  ] 15 tests.

You should see from the above unit tests results that the average cost time of the constant time median filter change little when the filer radius increate from 1 to 5.

Build on android

  1. Install adb tool with following command 
sudo apt update
sudo apt install android-tools-adb android-tools-fastboot

  1. Download android ndk 15c

  2. Install cmake 3.11

  3. Build the project by running the scripts/build_android.sh script and remember to modify the following parameter. 

BUILD_ANDROID_NDK_HOME=/home/ldpe2g/Android/android-ndk-r15c
DEPLOY_DIR=/data/local/tmp/ldp
CMAKE_PATH=cmake

  1. You can modify the USE_NEON=ON flag to enable or disable neon support. But you should see a huge performance degration on normal median filter when disable neon.

  2. After successfully building the project, the executable runUnitTests will be pushed to the folder ${DEPLOY_DIR} on your android mobile and run. If every thing goes right, you should see the similar outputs on your terminal as you build on linux. 

[==========] Running 15 tests from 1 test suite.
[----------] Global test environment set-up.
[----------] 15 tests from netTest
[ RUN      ] netTest.medianFilterFloatRaduis1SpeedTest

#################Normal Float MedianFilfer Radius 1 Average Cost time: 17.029ms

[       OK ] netTest.medianFilterFloatRaduis1SpeedTest (182 ms)
[ RUN      ] netTest.medianFilterUint16Raduis1SpeedTest

#################Normal Uint16 MedianFilfer Radius 1 Average Cost time: 7.6964ms

[       OK ] netTest.medianFilterUint16Raduis1SpeedTest (81 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis1SpeedTest

#################Constant time Uint16 MedianFilfer Radius 1 Average Cost time: 69.9052ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis1SpeedTest (703 ms)
[ RUN      ] netTest.medianFilterFloatRaduis2SpeedTest

#################Normal Float MedianFilfer Radius 2 Average Cost time: 57.2293ms

[       OK ] netTest.medianFilterFloatRaduis2SpeedTest (576 ms)
[ RUN      ] netTest.medianFilterUint16Raduis2SpeedTest

#################Normal Uint16 MedianFilfer Radius 2 Average Cost time: 25.823ms

[       OK ] netTest.medianFilterUint16Raduis2SpeedTest (261 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis2SpeedTest

#################Constant Time Uint16 MedianFilfer Radius 2 Average Cost time: 74.136ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis2SpeedTest (745 ms)
[ RUN      ] netTest.medianFilterFloatRaduis3SpeedTest

#################Normal Float MedianFilfer Radius 3 Average Cost time: 173.699ms

[       OK ] netTest.medianFilterFloatRaduis3SpeedTest (1742 ms)
[ RUN      ] netTest.medianFilterUint16Raduis3SpeedTest

#################Normal Uint16 MedianFilfer Radius 3 Average Cost time: 72.2515ms

[       OK ] netTest.medianFilterUint16Raduis3SpeedTest (726 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis3SpeedTest

#################Constant time Uint16 MedianFilfer Radius 3 Average Cost time: 75.1661ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis3SpeedTest (755 ms)
[ RUN      ] netTest.medianFilterFloatRaduis4SpeedTest

#################Normal Float MedianFilfer Radius 4 Average Cost time: 420.3ms

[       OK ] netTest.medianFilterFloatRaduis4SpeedTest (4208 ms)
[ RUN      ] netTest.medianFilterUint16Raduis4SpeedTest

#################Normal Uint16 MedianFilfer Radius 4 Average Cost time: 170.781ms

[       OK ] netTest.medianFilterUint16Raduis4SpeedTest (1711 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis4SpeedTest

#################Constant time Uint16 MedianFilfer Radius 4 Average Cost time: 76.0269ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis4SpeedTest (764 ms)
[ RUN      ] netTest.medianFilterFloatRaduis5SpeedTest

#################Normal Float MedianFilfer Radius 5 Average Cost time: 915.703ms

[       OK ] netTest.medianFilterFloatRaduis5SpeedTest (9162 ms)
[ RUN      ] netTest.medianFilterUint16Raduis5SpeedTest

#################Normal Uint16 MedianFilfer Radius 5 Average Cost time: 368.779ms

[       OK ] netTest.medianFilterUint16Raduis5SpeedTest (3691 ms)
[ RUN      ] netTest.constantTimeMedianFilterUint16Raduis5SpeedTest

#################Constant time Uint16 MedianFilfer Radius 5 Average Cost time: 75.8295ms

[       OK ] netTest.constantTimeMedianFilterUint16Raduis5SpeedTest (762 ms)
[----------] 15 tests from netTest (26070 ms total)

[----------] Global test environment tear-down
[==========] 15 tests from 1 test suite ran. (26070 ms total)
[  PASSED  ] 15 tests.

Run median filter on Linux

After successfully building the project on linux, you can then use the scripts/run_filters.sh to run the median filter with you own images. And remember to modify the following one or more parameters, then have fun :)! 

INPUT_PATH=${basepath}/images/jian20_salt_pepper_noise.jpg
FILTER_RADIUS=3
ITERATION_NUM=1
MEDIAN_FILTER_OUT_PATH_PREFIX=${basepath}/results/median_filter_radius${FILTER_RADIUS}_iter${ITERATION_NUM}

Run median filter on Android

After successfully building the project on android, you can then use the scripts/run_filters_android.sh to run the median filter with you own images on your android device. And remember to modify the following one or more parameters, then have fun :)! 

FILTER_RADIUS=3
ITERATION_NUM=1

INPUT_NAME=jian20_salt_pepper_noise.jpg

MEDIAN_FILTER_OUT_NAME=android_median_filter_radius${FILTER_RADIUS}_iter${ITERATION_NUM}

INPUT_PATH=${basepath}/images/${INPUT_NAME}
OUTPUT_PATH=${basepath}/results/

report

Recommend

About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK