Academia.eduAcademia.edu

Figure 1 – uploaded by Mustapha SAIDALLAH

See full PDFdownloadDownload figure
TABLE 1. DIFFERENT CHARACTERISTICS OF ACTIVE AND PASSIVE SENSORS THAT CAN BE USED IN VEHICLE DETECTION. Passive sensors such as camera are low-cost, easy to install, and provide a higher resolution with a wider view angle. Therefore, this paper is limited to Vision-Based object detection systems.

Table 1 DIFFERENT CHARACTERISTICS OF ACTIVE AND PASSIVE SENSORS THAT CAN BE USED IN VEHICLE DETECTION. Passive sensors such as camera are low-cost, easy to install, and provide a higher resolution with a wider view angle. Therefore, this paper is limited to Vision-Based object detection systems.

Related Figures (5)

Feature Extraction stage serves to extract the shape, silhouette or other specific features of the object. Then, the classifier uses those features to make a decision about the presence of a vehicle.
Feature Extraction stage serves to extract the shape, silhouette or other specific features of the object. Then, the classifier uses those features to make a decision about the presence of a vehicle.
Fig. 2. Image-preprocessing steps
Fig. 2. Image-preprocessing steps
According to viola and Jones [8], more than 180,000 potential characteristics can be generated for each 24 x 24 window. Thus, they proposed the integral image method to quickly calculate these descriptors. Fig. 3. Examples of local Haar wavelets.
According to viola and Jones [8], more than 180,000 potential characteristics can be generated for each 24 x 24 window. Thus, they proposed the integral image method to quickly calculate these descriptors. Fig. 3. Examples of local Haar wavelets.
The Detection rate (DR) and false alarm rate (FAR) are used to evaluate the performance of this descriptors. The best detector is the one that gives a high detection rate with the lowest false detection rate. According to the results obtained: the most pertinent features are Haar-like features, leading a DR of 90.1% and a FAR of 3%. Against the LBP filters with a DR of 87.9% and a FAR of 4.8%. Figure 5 illustrates vehicle detection on some input images of the database. Fig.5. Samples of vehicle detection based on Haar-Like (left), and LBP (right).
The Detection rate (DR) and false alarm rate (FAR) are used to evaluate the performance of this descriptors. The best detector is the one that gives a high detection rate with the lowest false detection rate. According to the results obtained: the most pertinent features are Haar-like features, leading a DR of 90.1% and a FAR of 3%. Against the LBP filters with a DR of 87.9% and a FAR of 4.8%. Figure 5 illustrates vehicle detection on some input images of the database. Fig.5. Samples of vehicle detection based on Haar-Like (left), and LBP (right).
Figure 6: Energy consumption results. The first graph is the result of Samsung Galaxy Tab A, The second graph is from sm-j100H. LBP is a Local Binary Pattern feature which offers a reduces complexity, comparing it with Haar-like features. Indeed, Haar- like features require larger calculations, memory and energy consuming.
Figure 6: Energy consumption results. The first graph is the result of Samsung Galaxy Tab A, The second graph is from sm-j100H. LBP is a Local Binary Pattern feature which offers a reduces complexity, comparing it with Haar-like features. Indeed, Haar- like features require larger calculations, memory and energy consuming.
Related topics:
Intelligent Transportation Systems
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved