Optimized outdoor parking system for smart cities using advanced saliency detection method and hybrid features extraction model

Figures & data

Table 1. Literature review of existing parking systems.

Reference	Objective	Technique	Accuracy
Alharbi et al. [7]	Automatic detection of traffic on streets and vital places in the cities	SPS, OCR	–
C. C. Huang et al. [24]	Introduced a vacant parking space detection system based on Bayesian hierarchical framework (BHF)	Image labelling for BHF	98.8%
Yusnita et al. [4]	Proposed an intelligent parking space detection system	Integrated image processing	–
		Approach
D. Delibaltov et al. [13]	Proposed an occupancy detection system based on the marked parking lot from lamp-post camera views	Segmentation algorithm and image processing techniques	80%
Al-Kharusi et al. [2]	Presented an optimized parking detection system	Image processing techniques	–
Qing Tian et al. [3]	Presented an intelligent parking management system	LPR (License Plate Recognition) through video streaming	95%
Deni et al. [31]	Detect vehicles on the side of the parking lot	Haar cascade classifier, YOLO	96.88%
Saharan et al. [33]	Parking occupancy and availability detection for motorists	Decision Tree, Neural Network, Random forest	–
Rizwan et al. [6]	Proposed secure parking framework	Support vector machine (SVM) for classification	73–91%.
Hoseok et al. [10]	Classification of parking slot using 480,000 images around view images	Convolution Neural Network	94.15%
M. Ahrnbom et al. [12]	Presented an efficient system to detect the presence of cars in parking lots	Machine learning algorithms SVM and Linear Regression with six quantized, gradient magnitude and colour space	98.68%
C. Badii et al. [19]	Presented a model to predict the number of available parking slots in city garages	Bayesian Regularized Neural Network	97.5%

Figure 1. Examples of images captured under different weather conditions: (a-e) from PUCPR; (f-j) from UCPR04 and (k-o) from UCPR05 respectively.

Figure 2. Proposed design of outdoor parking detection system.

Figure 3. Feature extraction techniques based on colour and texture features.

Table

Algorithm 1: Color_Feature_Extraction
Data: Parking Lot Dataset
Result: Feature vector F1 to F18 for all parking images
1. Apply Range on RGB and HSV color spaces and save six features (F1 to F6)
$R_I=Max(f_{Ij})-Min(f_{Ij})$
where $f_{Ij}$ is the value of the Ith color component of the image pixel j.
2. Apply Mean on RGB and HSV color spaces and save six features (F7 to F12)
${\mu }_I=\frac{1}{N}{\displaystyle \sum _{j=1}^N{f}_{Ij}}$
where N is the number of pixels in an image.
3. Apply Variance on RGB and HSV color spaces and save six features (F13 to F18)
${\sigma }_I=\sqrt{\frac{1}{N}{\displaystyle \sum _{j=1}^N{(f_{Ij}-{\mu }_I)}^2}}$
4. Store all the eighteen features F1 to F18 in the COLOR_FEATURES vector.

Table

Algorithm 2: Texture_Feature_Extraction
Data: Parking Lot Dataset
Result: Feature vector for all parking images
1. Read parking images in RGB color space.
2. Resize into $64\times 32$ size.
3. Apply extractHOGFeatures(Image)
4. Store them in TEXTURE_FEATURES vector.

Table

Algorithm 3: Proposed hybrid feature extraction Algorithm
Input: Parking Lot Dataset
Output: Feature vector for all parking images
For each class i = 1:2  % Two classes Vacant and Occupied if(i = =1) files = dir(type_1_path);  % Type_1 describes Vacant else files = dir(type_2_path);  % Type_2 describes Occupied end if end for if(i = =1)      % If Vacant for k = 3: length(files) file_path = fullfile(type_1_path,files(k).name); img = imread(file_path); features_1 = features_extraction_color(img); features_2 = features_extraction_texture(img); store_features([features_1’;features_2’],[1;0]) end for end if if(i = =2)       % If Occupied for k = 3: length(files) file_path = fullfile(type_2_path,files(k).name); img = imread(file_path); features_1 = features_extraction_color(img); features_2 = features_extraction_texture(img); end for end if Store all the features in the HYBRID_FEATURES vector.

Figure 4. (a) Original Cloudy image, (b) Histogram of original Image, (c) Image after CLAHE and (d) Histogram of Enhanced Image.

Figure 5. Arrangement of dataset challenges (a) original frame (b) Saliency (c) vacant or occupied frame.

Table 2. Comparison of proposed with the different existing models.

	PUCPR		UFPR04		UFPR05
	Accuracy	Processing time(ms)	Accuracy	Processing time(ms)	Accuracy	Processing time(ms)
KNN	92.57	12.35	93.65	11.85	92.45	12.59
ANN	98.74	7.35	98.99	7.83	98.85	7.72
NBC	91.57	13.74	90.37	13.87	90.47	14.02
Our model with SVM	99.93	10.35	99.89	10.58	99.87	10.79

Table 3. Comparison of SVM models on proposed system based on different Kernel Functions.

	PUCPR	UFPR04	UFPR05	Mean
“guassian”	99.93	99.89	99.87	99.89
“linear”	98.74	98.99	98.85	98.86
“polynomial”	91.57	90.37	90.47	90.80

Table 4. Experimental results of Proposed Hybrid Feature Extraction framework using SVM Classifier.

		Datasets
		PUCPR	UFPR04	UFPR05	Average (in %)
Performance Measures	Accuracy	99.93	99.89	99.87	99.89
	Sensitivity	99.28	99.56	99.84	99.56
	Specificity	99.76	99.84	99.84	99.81
	FPR	00.24	00.16	00.16	00.19
	FNR	00.72	00.44	00.16	00.44

Figure 6. ROC curves for the proposed model.

Table 5. Accuracy comparison of different feature extraction techniques with proposed hybrid feature extraction model.

	Feature extraction techniques
Datasets	Integral channel extraction by SVM	Integral channel extraction by LR	CNN correlated features	Proposed hybrid extraction Model
PUCPR	99.94	99.94	99.76	99.93
UFPR04	98.42	98.29	99.74	99.89
UFPR05	94.01	94.57	99.72	99.87
Mean	97.45	97.6	99.74	99.89

Table 6. Comparison of proposed with the other existing parking Methodologies.

Reference	Technique	Accuracy
C. C. Huang et al. [24]	Image labelling for BHF	98.8%
D. Delibaltov et al. [13]	Segmentation algorithm and image processing techniques	80%
Qing Tian et al. [3]	LPR (License Plate Recognition) through video streaming	95%
Deni et al. [31]	Haar cascade classifier, YOLO	96.88%
Rizwan et al. [6]	Support vector machine (SVM) for classification	73–91%
M. Ahrnbom et al. [12]	Machine learning algorithms SVM and Linear Regression with six quantized, gradient magnitude and colour space	98.68%
C. Badii et al. [19]	Bayesian Regularized Neural Network	97.5%
Present work	Hybrid feature extraction technique based on the Advanced Saliency Detection Method and Efficient Features Extraction Model	99.89%

Alharbi A, Ahlkias G, Yamin M, et al. Web based frame work for smart parking system. Int J Inf Technol. 2021;13:1495–1502.

 Google Scholar

Huang CC, Wang SJ. A hierarchical Bayesian generation framework for vacant parking space detection. IEEE Trans Circuits Syst Video Technol. 2011;20(12):1770–1785.

 Web of Science ®Google Scholar

Yusnita R, Norbaya F, Basharuddin N. Intelligent parking space detection system based on image processing. Int J Innov Manag Technol. 2012;3(3):232–235.

 Google Scholar

Delibaltov D, Wu W, Loce RP, et al. Parking lot occupancy determination from lamp-post camera images. In: IEEE conference on intelligent transportation systems, proceedings, ITSC, 2013. p. 2387–2392.

 Google Scholar

AI-Kharusi H, Al-Bahadly I. Intelligent parking management system based on image processing. World J Eng Technol. 2014;2:55–67.

 Google Scholar

Tian Q, Guo T, Qiao S. Design of intelligent parking management system based on license plate recognition. J Multimed. 2014;9(6):774–780.

 Google Scholar

Manase DK, Zainuddin Z, Syarif S, et al. Car detection in roadside parking for smart parking system based on image processing. In: 2020 international seminar on intelligent technology and its applications (ISITIA), IEEE, 2020. p. 194–198.

 Google Scholar

Saharan S, Kumar N, Bawa S. An efficient smart parking pricing system for smart city environment: a machine-learning based approach. Future Gener Comput Syst. 2020;106(1):622–640.

 Google Scholar

Rizwan M, Habib HA. Video analytics framework for automated parking. Tech J, Univ Eng Technol (UET) Taxila, Pakistan. 2015;20(4):87–94.

 Google Scholar

Do H, Kim J, Lee K, et al. Implementation of CNN based parking slot type classification using arounf view image. In: Proceedings of the 2020 IEEE international conference on consumer electronics (ICCE), 2020.

 Google Scholar

Ahrnbom M, Astrom K, Nilsson M. Fast classification of empty and occupied parking spaces using integral channel features. In: IEEE computer society conference on computer vision and pattern recognition workshops, 2016. p. 1609–1615.

 Google Scholar

Badii C, Nesi P, Paoli I. Predicting available parking slots on critical and regular services by exploiting a range of open data. IEEE Access. 2018;6:44059–44071.

 Web of Science ®Google Scholar