A Study of Feature Combination for Vehicle Detection Based on Image Processing

Table 9

Performance of the proposed ensembles for the left close/middle region.

(a)

PCA+V-HOG; + LG+𝑆+𝐺	Accuracy	Recall	Precision
RMM-SA	97.62	98.46	96.76
RMM-WA	98.42	98.68	98.16
DS-SA	97.50	98.26	96.72
DS-WA	98.64	98.33	98.96

(b)

PCA+V-HOG; + LG+𝑆	Accuracy	Recall	Precision
RMM-SA	97.78	97.88	97.68
RMM-WA	98.50	98.18	98.84
DS-SA	97.60	97.80	97.40
DS-WA	98.48	97.95	99.04

(c)

PCA+V-HOG+LG	Accuracy	Recall	Precision
RMM-SA	98.28	97.79	98.80
RMM-WA	98.22	97.58	98.92
DS-SA	98.28	97.75	98.84
DS-WA	98.26	97.52	99.04

(d)

PCA+LG+𝑆	Accuracy	Recall	Precision
RMM-SA	95.86	95.52	96.24
RMM-WA	98.36	98.10	98.64
DS-SA	95.30	95.10	95.52
DS-WA	98.30	97.75	98.88

(e)

PCA+LG	Accuracy	Recall	Precision
RMM-SA	97.46	97.18	97.76
RMM-WA	97.80	97.13	98.52
DS-SA	97.46	97.18	97.76
DS-WA	97.96	97.21	98.68