Scribd
Upload
96 views2 pages

Image Processing Algorithms

This document contains two tables summarizing the percentage of bad matching pixels and average time for different stereo matching algorithms on various images. The first table shows BM, SGBM, SAD and NCC algorithms had the lowest percentage of bad pixels, with BM being the fastest algorithm. The second table compares BM and SGBM on standard test images, with SGBM having a slightly higher percentage of bad pixels but being slower. A third source is cited comparing fps on a different image and hardware platform.

Uploaded by

Manoj Gurrala
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
96 views2 pages

Image Processing Algorithms

This document contains two tables summarizing the percentage of bad matching pixels and average time for different stereo matching algorithms on various images. The first table shows BM, SGBM, SAD and NCC algorithms had the lowest percentage of bad pixels, with BM being the fastest algorithm. The second table compares BM and SGBM on standard test images, with SGBM having a slightly higher percentage of bad pixels but being slower. A third source is cited comparing fps on a different image and hardware platform.

Uploaded by

Manoj Gurrala
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Below is a table that summarizes the percentage of bad matching pixels for BM,

SGBM, SAD and NCC algorithms.

Image 1 Road 1 Road 2


Average (s) FPS Average (s) FPS Average (s) FPS
BM 0.096 10.428 0.1125 8.8889 0.1137 8.7951
SGBM 1.038 0.963 1.9863 0.5034 1.9881 0.5030
SAD 34.694 0.029 60.8113 0.0164 61.7546 0.0162
NCC 29.937 0.033 58.0332 0.0172 57.0302 0.0175

Table 1. Characteristics of the most common image sets

Cone Teddy Tsukuba Venus


Size in pixels 450 x 375 450 x 375 384 x 288 434 x 383

Disparity 60 60 16 20
levels

Below is a table that summarizes the percentage of bad matching pixels for BM
and SGBM algorithms.

Avg. Time
OpenCV Cones Teddy Tsukuba Venus
BM 0.4450 0.4583 0.3556 0.4707 ~0.028 sec
SGBM 0.4943 0.4977 0.3923 0.4982 ~0.065 sec

Klaus et al. 2006 Segment-Based Stereo Matching Using Belief Propagation and
a Self-Adapting Dissimilarity Measure
Fps image Disparity levels Computational platform
0.07 384 x 288 32 16 AMD Athlon 64
2.21GHz

You might also like