Table 1.

Comparison of previous and the proposed methods. HOG, histogram of the oriented gradient; EM, expectation minimization; CSM, contour saliency map.

Category	Without Background Generation [7,8,9,15,16,17,18,19,20]	With Background Generation
		Not Adjusting the Parameters for Detection Based on Background Information [21,22,23,24,25,26,27,28,29]	Adjusting the Parameters for Detection Based on Background Information (Proposed Method)
Examples	-Motion + SIFT-based [2], HOG-based [7,8,9,15,16,17], geometric characteristics-based [15,17,18], Adaboost-based [19] and soft-label boosting-based [20] detection of object.	-Gaussian model-based [21,22,23,25], texture change-based [23], EM-based [26,27] and image averaging-based [28] background modeling and subtraction.	-The correct background image can be generated by image averaging, various filtering and erasing of the human area with adaptive determination of thresholds and parameters for the human detector.
		-CSM-based [21,24,25], CSM template matching-based [22], shape and appearance-based [26,27], spatiotemporal texture vectors-based [23], boosting framework-based [28] and particle filter and histogram-based [29] detection of object.
Advantages	-Can detect object without a background image.	-Can be applicable to various environmental conditions and can detect objects of various scales.	-Robust detection of the human area can be obtained by adaptively determining the thresholds and parameters for detection considering background information.
			-Does not require a training procedure to obtain the classifier of human detection.
Disadvantages	-Requires a pre-defined template or classifier for the human area that must be obtained through training.	-Performance degradation can occur if the intensity of the object is similar to the background.	-Additional procedure is required for obtaining correct background image.
	-Performance is influenced by various environmental factors, such as rain, snow and the amount of sunlight.	-The approach can produce erroneous background, including the image areas occupied by humans, if the image frames include the image areas occupied by motionless humans.
	-Requires significant processing time to detect the human area in the entire image by scanning.	-Parameters for the detector of the image areas occupied by humans are not adaptively determined based on background information.