Human Object Classification Using Dual Tree Complex Wavelet Transform and Zernike Moment - Transactions on Large-Scale Data-and Knowledge-Centered Systems - page 96

two feature set results as given in Table 3. This shows that combining two features
into one gives better results than use of single feature. Results shown in Tables 1-5,
clearly demonstrate that the proposed method gives better human object classification
results at higher levels of Dual tree complex wavelet transform in comparison to other
state-of-the-art methods proposed by Dalal and Triggs [7], Lu and Zheng [9], Renno
et al. [14] and Chen et al. [15].
We have also computed performance of the proposed method and all other state-of-
the-art methods in terms of three performance metrics - average classification accura-
cy, True positive rate (also known as Recall), and Predicted positive rate (also known
as Precision). For computation of these measures we have used four terms: TP (true
positive), TN (true negative), FP (false positive), FN (false negative), which are de-
fined as: TP are total number of images which are originally positive and detected as
positive images, TN are total number of images which are originally negative and
detected as negative images, FP are total number of images which are originally nega-
tive and detected as positive images, and FN are total number of images which are
originally positive and detected as negative images.

Table 3. Confusion matrix for the proposed method using combination of Dual tree complex
wavelet transform (DTCWT) coefficients at multilevel and Zernike moment coefficients as a
feature set

Combination of DTCWT coefficients
(level - 1) and Zernike moment
coefficient as a feature set

Combination of DTCWT coefficients
(level - 2) and Zernike moment
coefficient as a feature set

Predicted classes Predicted classes

True classes True Classes

Human Non-human Human Non-human

Human 95 5 Human 95 5

Non-human 10 90 Non-human 7 93

Combination of DTCWT coefficients
(level - 3) and Zernike moment
coefficient as a feature set

Combination of DTCWT coefficients
(level - 4) and Zernike moment
coefficient as a feature set

Predicted classes Predicted classes

True classes True classes

Human Non-human Human Non-human

Human 96 4 Human 96 4

Non-human 7 93 Non-human 5 95

Combination of DTCWT coefficients
(level - 5) and Zernike moment
coefficient as a feature set

Combination of DTCWT coefficients
(level - 6) and Zernike moment
coefficient as a feature set

Predicted classes
Human Non-human Human Non-human
Human 99 1 Human 99 1
Non-human 3 97 Non-human 2 98
Combination of DTCWT coefficients (level - 7) and Zernike moment coefficient
as a feature set

Predicted classes

True classes True classes

Predicted classes

True classes

Human Non-human

Human 100 0

Non-human 2 98