Skip to main content

Table 5 Performance comparison of the proposed algorithms with the literature

From: Multiscale texture retrieval based on low-dimensional and rotation-invariant features of curvelet transform

 

Algorithm

Feature size

Comput. complexity of feature

Class. method

Better than the proposed algorithms

Comparable with the proposed algorithms

Worse than the proposed algorithms

1

F μσ ′

84

Low

NN

-

-

a,b,c,d,e,f

2

F [13]

84

Low

NN

-

-

a,b,c,d,e,f

3

LBP 8 , 1 riu 2 / VAR 8 , 1 [20]

160

Medium

NN

b

-

c,d

4

LBPV 8 , 1 u 2 GM PD 2 [20]

227

Medium

NN

-

-

b,c,d

5

LBP 24 , 3 riu 2 / VAR 24 , 3 [20]

416

Medium

NN

b

c,d

-

6

LBP 8 , 1 u 2 GM ES [20]

451

Medium

NN

-

-

b,c,d

7

KLD [15]

840

Medium

NN

-

-

a,e

8

LBPV 24 , 3 u 2 GM PD 2 [20]

2,211

High

NN

b,c,d

-

-

9

LBP 24 , 3 u 2 GM ES [20]

13,251

High

NN

b,c,d

-

-

10

HSR + LSR [16] RIFT

4,000

High

SVM

f

-

a,e

11

HSR + LSR [16] SIFT

5,120

High

SVM

a,e,f

-

-

  1. The table does not include the proposed algorithms and is ordered in the increasing complexity and feature size order (top to bottom). The databases are represented with as follows: a, Brodatz; b, TC10, c, TC12-horizon; d, TC12-t184; e, KTH-TIPS; f, UIUCTex. The italicized letters indicate the databases where the proposed methods are superior.