555
©IJETIE 2019
3D OBJECT RECOGNITION AND
DETECTION USING SURF MAPPING
R.Kabilana, Dr.R.Ravib,S.Suhirthac, M.Sankara Gomathid, S.Sofiae, Department ofECE, Francis Xavier Engineering College, Tirunelvelia,b,c,d,e
Assistant Professor, Department ofECEa, Professor Department ofITa , Final year students c,d,e, Francis Xavier Engineering College, Tirunelveli
[email protected]a,[email protected]b, [email protected]c, [email protected]d, [email protected]e
ABSTRACT
The goal of the project is to recognize objects and track them. In this project algorithm uses a combination of Object detection and image-based template matching to track the targets. Object detection is determined by temporal differencing and template matching is done only on the locations as guided by the Object detection stage to provide a robust target-tracking method. Results show no false object recognition in all tested images, perfect tracking for the synthetic images and 98% tracked rate on the real images.
Keywords: Foreground and Background Separation Edge Detection, Histogram Operation, Edge Detection, Binary Operation, Thresholding.
1. INTRODUCTION
There are two main goals to our project: to correctly recognize moving objects of interest, and to track those moving objects throughout their life spans. The problem facing the first task is noise: some of the moving pixels in the image may be due to camera Object or noisy background that is not of interest.
We solve this problem using size constraint to filter out movements that are too small to be part of an object of interest.
The problem facing the second task is how to correctly identify the objects of interest in all the image frames. Using Object detection alone to track the targets provides a poor method due to the noisy movements. Template matching alone is not robust against in change object orientation and tends to drift off when targets are in Object. In multitask learning was used in
multiclass object detection to share features between objects from different categories and decrease the amount of training data. The domain adaptation technique was utilized to transfer useful knowledge from another domain. They utilized some objects from Google’s 3-D warehouse to train an object detection system for 3-D point clouds collected by robots navigating through both urban and indoor environments. Many algorithms and models have also been proposed for imbalanced learning. The two models of evolutionary fuzzy ARTMAP neural networks were proposed to deal with the imbalanced data set problems. A critical review of the state-of the-art technologies and the current assessment metrics used to evaluate learning performance under the imbalanced learning scenario. Hence, we use a combination of the two by
performing template matching only on the pixels that are moving. Results have shown this to be a robust method in both recognizing the objects as well as tracking them.
2. EXISTING METHOD
Object detection and object tracking is a very rich research area in computer vision. The main issues that make this research area difficult are:
Computational Expense. If an algorithm for detecting Objects and tracking objects is to be applied to real-time applications, then it needs to be computationally inexpensive so that a modern PC has enough power to run it. Yet many algorithms in this research area are very computationally expensive since they require computing values for each pixel in each image frame.
• Moving Background Rejection. The algorithm also needs to be able to reject moving background such as a swaying tree branch and not mistakenly recognizes it as a moving object of interest. Misclassification can easily occur if the area of moving background is large compared to the objects of interest, or if the speed of moving objects is as slow as the background.
• Tracking Through Occlusion. Many algorithms have devised ways of becoming robust against small occlusions of interested objects, but most algorithms still fail to track the object if it is occluded for a long period of time.
• Modeling Targets of Interest. Many algorithms use a reasonably detailed model of the targets in objects detection and consequently require a large number of pixels on target in order to detect and track them properly. This is a problem for real-world applications where it is frequently impossible to obtain a large number of pixels on target.
• Adapting to Illumination Variation. Real-world applications will inevitably have variation in scene illumination that a Object detection algorithm needs to
cope with. Yet if an algorithm is purely an intensity- based method then it will fail under illumination variation.
• Analyzing Object. After objects have been correctly classified and tracked, an algorithm may want to analyze the object Objects such as the gait of moving human, the speed of car (is it speeding or not?) etc…
This could be difficult especially for non-rigid objects such as human if the object view is not in the right perspective for the algorithm.
• Adapting to Camera Object. Detecting moving entities from mobile camera video streams still remains a challenge in this research area.
2.1. TEMPORAL CONSISTENCY
The idea is that the true object should be moving long enough or it will be just the transient background movement.
The background might be falsely classified as the object. If this occurs, it should later be removed from the template list.
Approach for updating the template list.
1. Record all potential target Pn(i) = Rn(i)from some initial frame. (Rn(i) is the Object region ith in the nth frame)
2. Record classification hypothesis X(i) = { ID(Pn(i) } where ID(x) is the identification metric operator, in this case we use image based metric to identify whether the object is a human or a car.
3. For new frames, each previous Object region Pn- 1(i) is matched to the spatially closest Object region Rn(j).
4. Pn-1(i) that does not find the match will be removed from the template list.
5. Rn(j) that does not match any previous Pn-I(i) will be added to the template list.
X(i) = {X(i)} U { ID(Pn(i)) }
2.2. TARGET CLASSIFICATION
A classification histogram, using ID () based on dispersions, is computed for each Object region at each time, i.e. tclass. Every instant after tclass, the peak of histogram of each Object region is used to classify the target. The object will be reclassified every tclass. This is to prevent misclassification of the object by exploiting the fact that if something affects the object, e.g. occlusion, or object coming from the edge, it will not last for a long time. Eventually, when that thing is gone, the histogram of the correct class will outnumber the histogram of the misclassification and the object will be correctly classified.
This also gives the tracking more robust to background clutter such as leaves blowing. Again, this false object movement is transient will not last longer than tclass.
3. PROPOSED METHOD
3.1. EXPERIMENTATION AND PROGRAM DEVELOPMENT
We design our experiments to solve the issues that we encountered. The main issues are: dealing with the false positives in the object recognition stage due to camera movements or noisy background; choosing threshold and range parameters to segment the moving regions from the background and combining overlapping regions to create the template of the objects of interest; template matching to give the best tracking results.
3.2. EXPERIMENT DESIGN
Our first goal is to eliminate the false positives due to noisy movements in the Object image. We accomplish this by:
1. Use size filter to eliminate the movements that are too small to be part of an object of interest.
2. Test it with synthetic images such as syn1car.vx and syn2car.vx and experiment with different size filter parameter to minimize the number of false positives.
3. Next find a relationship between the size of the filter to the size of the image so that the program can determine what filter size to use automatically.
Our second goal is to create the template. We accomplish this by:
• Use region growing to group moving pixels in Object image into Object regions and use size filter to eliminate regions that are too large and likely to belong to background.
• Test with the synthetic images and experiment with different range parameter to obtain best sets of regions as well as different size filter as to minimize the amount of background drifting on to the object regions.
• Group regions together that belong to same object by using equivalent table.
• Test again. Repeat from step 2 until the right amount of objects are recognized, then create the template for each object.
Since the synthetic images have no noise and can easily be segmented, they present no problem to our first and second goal hence we also test our algorithm with small real images such as truck2.vx and truck1.vx so to give a more accurate picture of the performance of our algorithm. We tested our program with both synthetics image frame and real image frame. The synthetic images were created in Photoshop. We created them to be small and not complicated for testing our program in the initial step. The moving objects have intensity 255 and are clearly different from the background. Our program performs very well on synthetic images. It can track both one and two moving objects correctly. Synthetic images used in this experiment are syn1car.vx and syn2car.vx.
We obtained the real image sequences from the database of ee547. The images used are truck1.vx, truck2.vx, and van.vx.
In real images, the moving object and background do not clearly differ as in synthetics image case. The sequences are also longer. The real images give more accurate performance test for our program.
We also compare our tracking program with Vision X, tracking command vtrack and list the feature of each tracker as follows.
➢ Vtrack
• A user must specify the location to be tracked. The program track 9 x 9 windows centered at the location specified by the user.
• Tracking performance is very high. Each object specified is correctly tracked through the whole frame.
➢ Our Tracking
• Track moving object automatically.
• Detect moving objects, and segment them.
• Tracking performance is degraded when moving object and background have little intensity difference.
Fig 3.1 Block diagram A. Vision System:
For many vision-based systems, it is important to detect a moving object automatically Image processing, analysis, and machine vision represent an exciting and dynamic part of cognitive and computer science. From vision system pc should installed frame grabber card which is support to Mobile camera and it should have fast processor for capturing frame with snapshot.
B. Object Detection in 2D:
There are many algorithms for detecting objects in a data stream. One such algorithm that was researched and tested for this project is described in [HLGT03]. Videos containing background objects under various conditions pertaining to lighting and movement were processed and moving objects were segmented from the foreground based on classification under Bayesian rules
C. Adaptive Background Mixture Model
Background subtraction is processed on the scene to segment the foreground objects. The background model is based on a mixture of gaussian distributions, and different weights are used to represent the temporal portions of colors in the scene. When marking a pixel, weights are applied to each gaussian, and the one with the most probable outcome is used to mark the pixel.
Kalman Filter Another important aspect of the project was handling noise output from the sensor. All data output from the Kinect is subject to noise, and for this reason it was necessary to process the data through a filter.
D. 3D Visualization and Augmented Tracking with PCL The Point Cloud Library (from here on referred to as PCL) was used PCL is an industry standard library for processing 3-dimensional point cloud data. It was chosen as the 3D rendering engine for the project due to its integration with ROS. As in the optical tracker implementation, there were significant problems in using the API due to changes in the latest stable release and ROS not being up to date at the time of this project’s implementation. PCL features many high-level implementations of general point cloud processing applications, and provides class-based implementations of algorithms related to 3D processing. PCL’s high-level visualization toolkit, discussed below, was used to incorporate 3D data from the Kinect into the project. The following subsections aim to describe further the use of PCL in the project to render 3D point cloud data.
4. RESULTS
The object recognition problem in sequence truck1.vx is a consequence of poor segmentation of the object which fails to recognize the whole car as the object. However, the tracking algorithm is not severely affected since it is guided by the Object detection stage resulting in a tracking rate of more than 90 % of the time. Our program is reasonably efficient with execution time less than 40 s on 30 image sequences of 400 x 60 pixels. The execution time is increased to over 4 minutes on the same sequences when the template matching was done on all pixels in the image and not just the moving pixels. The tracking results were also virtually identical. Though our
algorithm does not seem to detect many false positives/negatives, the test images did not contain too many noisy movements. This is partly due to the small amount of noisy movements that our size filters manage to get rid of, and partly due to a constant appearance in the objects throughout their life spans. Moreover, our algorithm will not be able to track new objects of interest that are introduced in the middle of the sequence because new templates are not created once the program is underway, the templates are only updated as the sequence runs through. Future work could look to implement this.
Figure 4.1 object detection Figure 4.3 original, RGB, restored image
Figure 4.2 3D image
Figure 4.4 Histogram graph
5. CONCLUSIONS
For this project, we achieve with an object-tracking program that can automatically track multiple objects. We add intelligence to our program to track moving object automatically without being specified by users. It detects the Object, segments moving objects, and then tracks them in the frames following the first two frames. Tracking accuracy is quite good based on the result that moving objects are correctly tracked through the whole sequence.
REFERENCES
[1]. K. Lai, L. Bo, X. Ren, and D. Fox, “Detection-based object labeling in 3D scenes,” in Proc. 29th IEEE Int.
Conf. Robot. Autom., Saint Paul, MN, USA, May 2012, pp. 1330–1337.
[2]. N. Muthukumaran, 'Analyzing Throughput of MANET with Reduced Packet Loss', Wireless Personal Communications, Vol. 97, No. 1, pp. 565-578, November 2017.
[3]. N. Muthukumaran and R. Ravi, 'VLSI Implementations of Compressive Image Acquisition using Block Based Compression Algorithm', The International Arab Journal of Information Technology, vol. 12, no. 4, pp.
333-339, July 2015.
[4]. A.Dinesh Kumar, R.Pandi Selvam and K.Sathish Kumar “Review on Prediction Algorithms in Educational Data Mining” International Journal of Pure and Applied Mathematics(IJPAM)-Vol:118- No.:8,pp.531-537,2018. ISSN:1311-8080.-Scopus Indexed.
[5]. R.Pandi Selvam and V.Palanisamy “An Efficient Cluster-Based Multi-Source Multicast Routing Protocol in Mobile Ad-hoc Networks”, IEEE International Conference on Communication Control and Computing Technologies, Syed Ammal Engineering College, Ramanathapuram on 7-9th October 2010. pp.414-419. ISBN: 978-1-4244-7768-5.
[6]. R.Pandi Selvam and V.Palanisamy “An Optimized Cluster-Based Approach for Multi-Source Multicast Routing Protocol in Mobile Ad-hoc Networks with Differential Evolution”, IEEE International
Conference on Pattern Recognition, Informatics and Medical Engineering ”, Periyar University, Salem on 21st – 23rd March 2012. pp. 115-120. ISBN: 978-1- 4673-1039-0.
[7]. B. Manoj Kumar and N. Muthukumaran, 'Design of Low power high Speed CASCADED Double Tail Comparator', International Journal of Advanced Research in Biology Engineering Science and Technology, Vol. 2, No. 4, pp.18-22, June 2016.
[8]. P. Venkateswari, E. Jebitha Steffy, Dr. N.
Muthukumaran, 'License Plate cognizance by Ocular Character Perception', International Research Journal of Engineering and Technology, Vol. 5, No. 2, pp. 536- 542, February 2018.
[9]. N. Muthukumaran, Mrs R.Sonya, Dr. Rajashekhara and V. Chitra, 'Computation of Optimum ATC Using Generator Participation Factor in Deregulated System', International Journal of Advanced Research Trends in Engineering and Technology, Vol. 4, No. 1, pp. 8-11, January 2017.
[10]. Ms. A. Aruna, Ms.Y.Bibisha Mol, Ms.G.Delcy, Dr.
N. Muthukumaran, 'Arduino Powered Obstacles Avoidance for Visually Impaired Person', Asian Journal of Applied Science and Technology, Vol. 2, No. 2, pp. 101-106, April 2018.
[11]. B. Renuka, B. Sivaranjani, A. Maha Lakshmi, Dr. N.
Muthukumaran, 'Automatic Enemy Detecting Defense Robot by using Face Detection Technique', Asian Journal of Applied Science and Technology, Vol. 2, No. 2, pp. 495-501, April 2018.
[12]. Ms. Mary Varsha Peter, Ms. V. Priya, Ms. H.
Petchammal, Dr. N. Muthukumaran, 'Finger Print Based Smart Voting System', Asian Journal of Applied Science and Technology, Vol. 2, No. 2, pp. 357-361, April 2018.
[13]. R. Sudhashree, N. Muthukumaran, 'Analysis of Low Complexity Memory Footprint Reduction for Delay and Area Efficient Realization of 2 D FIR Filters', International Journal of Applied Engineering Research, Vol. 10, No. 20, pp. 16101-16105, 2015.
[14]. F.M. Aiysha Farzana, Abhinaya. M. K, Mrs.Friska, Dr. N. Muthukumaran, 'Design of Button Antenna for Wireless Body Network using HFSS', Indo-Iranian
Journal of Scientific Research, Vol. 3, No. 1, pp. 48- 54, January 2019.
[15]. F.M. Aiysha Farzana, Hameedhul Arshadh. A, Sara Safreen. M, Dr. N. Muthukumaran, 'Design and Analysis for Removing Salt and Pepper Noise in Image Processing', Indo-Iranian Journal of Scientific Research, Vol. 3, No. 1, pp. 42-47, January 2019.
[16]. A.Srinithi, E.Sumathi, K.Sushmithawathi, M.Vaishnavi, Dr. N. Muthukumaran, 'An Embedded Based Integrated Flood Forecasting through HAM Communication', Asian Journal of Applied Science and Technology, Vol. 3, No. 1, pp. 63-67, January 2019.
[17]. F.M.Aiysha Farzana, Hameedhul Arshadh. A, Ganesan. J, Dr. N. Muthukumaran, 'High Performance VLSI Architecture for Advanced QPSK Modems', Asian Journal of Applied Science and Technology, Vol. 3, No. 1, pp. 45-49, January 2019.
[18]. Banumathi.A, Banupriya.A, Niranjana.R, Jayaraman.G, Dr. N. Muthukumaran, 'Advanced Illumination Measurement System in Highways', Asian Journal of Applied Science and Technology, Vol. 3, No. 1, pp. 39-44, January 2019.
[19]. Lakshmanaprabu S.K, Sachi Nandan Mohanty, Sheeba Rani S, Sujatha Krishnamoorthy, Uthayakumar J, K. Shankar, “Online clinical decision support system using optimal deep neural networks”, Applied Soft Computing Journal, Volume 81, Page(s): 1-10, August 2019. (Impact Factor: 4.873) [20]. Mohamed Elhoseny, Gui-Bin Bian, SK.
Lakshmanaprabu, K. Shankar, Amit Kumar Singh, Wanqing Wu, “Effective Features to Classify Ovarian Cancer Data in Internet of Medical Things, Computer Networks”, Volume 159, Pages 147-156, August 2019. (Impact Factor: 3.030)
[21]. Mrs. S. Murine Sharmili, Dr. N. Muthukumaran, 'Performance Analysis of Elevation & Building Contours Image using K-Mean Clustering with Mathematical Morphology and SVM', Asian Journal of Applied Science and Technology, Vol. 2, No. 2, pp. 80- 85, April 2018.
[22]. N. Muthukumaran and R. Ravi, 'Hardware Implementation of Architecture Techniques for Fast
Efficient loss less Image Compression System', Wireless Personal Communications, Volume. 90, No.
3, pp. 1291-1315, October 2016.
[23]. N. Muthukumaran and R. Ravi, 'The Performance Analysis of Fast Efficient Lossless Satellite Image Compression and Decompression for Wavelet Based Algorithm', Wireless Personal Communications, Volume. 81, No. 2, pp. 839-859, March 2015.
[24]. I. Dryanovski, R. G. Valenti, and J. Xiao, “Fast visual odometry and mapping from RGB-D data,” in Proc.
30th IEEE Int. Conf. Robot. Autom., Karlsruhe, Germany, May 2013, pp. 2305–2310.
[25]. A. Anand, H. S. Koppula, T. Joachims, and A.
Saxena, “Contextually guided semantic labeling and search for three-dimensional point clouds,” Int. J.
Robot. Res., vol. 32, no. 1, pp. 19–34, 2013.
