Keynotes & Invited
REGISTRATION until 30 June 2017 @ http://bit.ly/2quaQo3
|19 July 2017 (Wednesday)
@ Auditorium 1, Level 3 (APU new campus)
Asia Pacific University of Technology and Innovation (APU)
Jalan Teknologi 5, Technology Park Malaysia, Bukit Jalil,
Kuala Lumpur, Malaysia
IEEE Signal Processing
Society Malaysia Chapter is jointly organizing the 8th Symposium
on Image Processing, Image Analysis and Real-Time Imaging (IPIARTI
2017) with the IEEE APU Student Branch and APU Graduate Student Council
This FREE annual event, open to all members and non-members of the IEEE, is organized with the objectives of:
· to bring the university and industry community together to share and discuss the latest trends in image processing, analysis and real-time implementation, and
· to promote the IEEE Signal Processing Society Malaysia Chapter to the academic and industry community in Malaysia as a forum for professional networking and advancement.
IEEE SPS Malaysia Chapter /
Asia Pacific University of Technology &
Local Arrangement :
IEEE APU Student Branch (IEEE APU SB)
APU Graduate Student Council (APU GSC)
Syed Abdul Rahman Syed Abu Bakar
Universiti Teknologi Malaysia (UTM)
Mohammad Faizal Ahmad Fauzi
Multimedia University (MMU)
Nor’aini Abdul Jalil
Syed Khaleel Ahmed
Universiti Tenaga Nasional (UNITEN)
Vijanth Sagayan Asirvadam
Universiti Teknologi PETRONAS (UTP)
Universiti Malaysia Pahang (UMP)
Hezerul Abdul Karim
Multimedia University (MMU)
Kushsairy Abdul Kadir
Universiti Kuala Lumpur (UniKL)
Wong Kok Sheikh
Monash University (MONASH)
Mohd Norzali Haji Mohd
Universiti Tun Hussein Onn Malaysia
ABSTRACT Submission FOR TECHNICAL PRESENTATIONS
|Back to Top
Professor Dr M Iqbal Saripan
Deputy Vice Chancellor (Academic & International)
Universiti Putra Malaysia (UPM)
Synopsis: -will be updated soon -
-will be updated soon -
Dr Sophea Prum
MIMOS Berhad, Malaysia
and Its Applications:
Synopsis: Nowadays, smartphone, digital camera and CCTV play a vital role in our personal usage, business and nationwide security. A smartphone can be used not only as a personal camera but also as a personal scanner allowing to scan any document instantly. In this use-case, document image processing systems are needed in order to extract relevant information.
security purpose, CCTV
been installed almost
will covert two systems. The first system focuses on document image
processing. It consists
in instantly extracting relevant information from the document images
captured by smartphone or any other similar devices. The second system
focuses on video processing for vehicle category/make
and model recognition.
Dr. Sophea Prum is a research engineer at Mimos (National R&D center of ICT) in Malaysia since August 2015. She was researcher and assistant lecturer at Laboratory of Information, Image and Interaction (L3i), University of La Rochelle from 2012 to 2015. She received her engineering degree in computer science from Institute of Technology of Cambodia (ITC) in 2008. She received her Master and PhD degrees respectively in 2009 and 2013 from University of La Rochelle. Her research focuses on image/video processing, document image processing, handwriting recognition and machine learning.
Dr Muhammad Khalis Bin Abdul Karim
Senior Medical Physicist,
National Cancer Institute, Putrajaya
Synopsis: According to the World Cancer Report 2014, cancer is a major cause of mortality where there were 8 million cancer related deaths in 2012, affecting the population in all countries of the world. However, it is important to state that our aim in healthcare is to prolong the survival rate among cancer patients through good clinical management. Radiology services in healthcare is known for its ability to provide diagnosis of cancers or tumours and help in managing cancer patients. However, since there is "a new kid on the block" known as radiogenomic, the diagnosis for tumours staging and prognosis, an even for treatment, is stepping into a new era. A radiogenomic imaging is a combination technique of radiological diagnosis and genomic assessment. Therefore, its flexibility in integrating both aspects, investigating the relationship between imaging features and gene expression, brings us from X-ray images to DNA analysis. Therefore, radiogenomic in future would be the most powerful tool for molecular assessment of tumour staging and diagnosis. This talk will introduce the imaging modality necessary to develop radiogenomic features, which involves the principles of MRI and CT scans, and the necessities of laboratory instruments in hospitals or research institutions. The talk will also briefly introduce work done by other researchers aroundthe world in the race to develop proper techniques for radiogenomic.
Muhammad Khalis Bin Abdul Karim obtained his B.Sc (Hons) in Imaging
Diagnostic and Radiotherapy from Universiti Kebangsaan Malaysia (UKM)
in 2008 and M.Sc in Physics from Universiti Teknologi Malaysia (2013).
In 2017, he completed his Ph.D in Physics at the Universiti Teknologi
Malaysia (UTM). He started his career as an Executive with a private
hospital, Mahkota Medical Centre in 2008 before joining Jabatan
Kesihatan Negeri Johor as a Medical Physicist in year 2009. Currently,
he is a Senior Medical Physicist and the Chief Physicist of Radiology
services in National Cancer Institute. Some of his achievements
include Excellent Service Award by the Federal Government
to recognize his achievements, 1st place for Best Presentation in
International Conference on Engineering, Science and Humanity (ICGESH)
in 2016, 1st place in Three Minute Thesis Competition (UTM)
Chancellor’s Award for his Ph.D project and Best Postgraduate Student
in the Physics Doctoral programme.
|INVITED TALK #1
Herdawatie Abdul Kadir
and Mohd Rizal Arshad
Universiti Tun Hussein Onn Malaysia
Batu Pahat, Johor
|Cooperative Simultaneous Localization
and Mapping: Enhanced Feature Detection Method for Ocean Observation
Navigation in an ocean environment with few static features and dynamic water background is an adventurous field to be explored by multi-agent system. This is because of its non-uniform availability of measurement on the ocean surface since the spatial feature distribution is greatly varied. Thus, it is desirable to design a cooperative localisation and mapping framework that is capable to handle spurious detection, reduce the localisation uncertainty of an agent and achieve fast and good decision. The main objective of this research is to design a cooperative simultaneous localisation and mapping method for multi blimp system involving the dynamic water surface as the background and small flock consensus as the group decision method. A new cooperative framework for the multi blimp system consisting of three blimps and buoys was developed and designed for this purpose. The simultaneous localisation and mapping were designed by integrating three methods which are the Extended Kalman Filter, the enhanced Scale Invariant Feature Transform and Received Signal Strength Indicator to improve the data association process. The group perception of direction based on small flock of animal consensus was taken into the data association process. It was discovered that this cooperative simultaneous localisation and mapping was able to reduce the number of feature points and detect the desired features in clear and dark water environments. In addition, based on cooperative benchmarking, this method was able to achieve faster consensus to up to 8.3 % and 42 % than the scale free model and klemm-eguilez model respectively. On top of these, its heading accuracy was found to be more accurate to up to 30 % and 76 % than the scale free model and klemm-eguilez model respectively. Overall, the proposed approach has achieved its prominent results and it is proven to be significantly reliable and applicable to be implemented in the ocean observation monitoring system.
|INVITED TALK #2
Md. Baharul Islam
Asia Pacific University of Technology & Innovation,
Technology Park Malaysia, Kuala Lumpur
|WS3D-VidR: Warping-based Stereoscopic
3D Video Retargeting with Depth Remapping
Due to the recent availability of stereoscopic display devices and cameras/lens, there is a growing demand for stereoscopic video retargeting methods that can automatically resize a stereo video to fit displays of different sizes. While depth plays an important role in influencing the user experience, to our best knowledge, none of the state-of-the-arts stereo video retargeting methods aims to enhance the depth of the retargeted video. We propose a warping-based approach that resizes and re-maps the depth of a stereoscopic video simultaneously to produce a better 3D experience. Firstly, our method computes the significance map for each stereoscopic video frame. It then performs volume warping using non-homogeneous scaling optimization to resize the stereoscopic video. During the warping optimization, a depth remapping constraint is used to remap the depth and significance content preservation constraint is applied to preserve the important content. Due to the significance content preservation across the video volume and the nature of the non-homogeneous scaling used for volume warping, motion consistency can be preserved without explicit temporal constraint. Experimental results demonstrate the effectiveness of our method in preserving the important content, ensuring motion consistency and enhancing the depth perception of the retargeted video.
|INVITED TALK #3
N. Dahlia Yusoff ,
S. Khairunniza-Bejo and Hazreen Harith
Universiti Putra Malaysia,
3D Shape Reconstruction of Fresh Fruit Bunch (FFB)
The reconstruction of 3D shape Fresh Fruit Bunch was carried using Poisson Surface Reconstruction (PSR) application, based on Delaunay triangular mesh generation techniques. The Delaunay triangulation explored the neighbourhood of a sample point cloud in all relevant directions in a way that even accommodates non-uniform samplings. Five experimental runs were conducted using MeshLab with refinement and coarsening of simplicial FFB shape. Results showed that the method was best agreed with a set of noisy, non-uniform observations, and it has been demonstrated that this approach can robustly recover fine detail from noisy real-world scan. Furthermore, the results showed some guarantees on the resulting mesh which enable the user to control the size and shape of mesh elements and the accuracy of the surface approximation. The reconstructed 3D models of Fresh Fruit Bunch are sufficiently accurate and realistic for 3D visualization in various applications.
|INVITED TALK #4
Rostam Affendi Hamzah,
M. Saad Hamid, Ahmad Fauzan Kadmin, S. Fakhar Abd Ghani, S. Salam and T. M. F. T. Wook
Universiti Teknikal Malaysia Melaka,
Durian Tunggal, Melaka
Improvement of Stereo Matching Algorithm for Low Texture Regions
The aim of stereo matching algorithm is to obtain the information of depth or distance to an input image. This is done by finding the matching pixel between two images at different viewpoints. The two-dimensional mapping of matching pixels is known as disparity map. This research topic has been studied for decades, which the depth from stereo remains a long-standing challenge. Several factors that make computations of stereo matching algorithm are challenging such as complex scenes, radiometric changes and repetition textures. The applications of depth stereopsis are intelligent vehicles, autonomous robotics and augmented reality. Stereo matching algorithms can be categorized into global and local methods. Global methods perform a matching process using global energy or a probability function over the whole image. The global methods involve high computational complexity and slow implementation. Therefore, it is not suitable for real-time applications. However, local methods solve the matching problem via a local analysis and aggregating matching costs over a support region at each pixel in the images. The local methods deliver fast execution and low computational requirement. The main challenge of stereo matching algorithm is to find the corresponding pixels in the low texture regions. These regions contain plain color pixels which make the corresponding process unable to determine the best matching pixels. Hence, this abstract proposes a new local-based stereo matching algorithm to improve the low texture regions which involves four stages. First, the matching cost function is developed using a combination of Absolute Differences (AD), Gradient Matching (GM) and Census Transform (CT). Second, the new edge preserving filter is proposed at cost aggregation stage. This filter is known as iterative guided filter which is able to increase the efficiency of preserving the object edges. Then, the optimization step uses a Winner-Take-All (WTA) strategy. The WTA strategy absorbs the minimal aggregated corresponding value for each valid pixel. Finally, the post-processing stage which is to refine the final disparity map. The unwanted and invalid pixels are still occurring at the occlusion and untextured areas. These unwanted pixels will be detected by Left-Right (LR) consistency checking process. Then, the fill-in process is carried out to replace the invalid pixels with a valid minimum pixel value. The disparity refinement step consists of implementing the weighted bilateral filter to remove the remaining noise which usually occurs during the fill-in process. The undirected graph segmentation and least square plane fitting process are used at the final step to recover the low texture regions on the final disparity map. Based on the experimental results of standard benchmarking from the Middlebury dataset, the proposed algorithm is able to reduce the error and increase the accuracy against the low textured areas with 36.01% of noise reduction. The proposed algorithm also produces good results on the real stereo images from the KITTI dataset.
|INVITED TALK #5
Mohd Norzali Haji Mohd and Muhammad Mahadi Abdul Jamil
Universiti Tun Hussein Onn Malaysia (UTHM),
Batu Pahat, Johor
An Optimized Low Computational Algorithm for Human Fall Detection from Depth Images Based on Support Vector Machine Classification
Systems developed to classify human activities to identify unintentional falls are highly demanding and play an important role in our daily life. Human falls are the main obstacle for elderly people to live independently and it is also a major health concern due to aging population. Different approaches are used to develop human fall detection systems for elderly and people with special needs. The three basic approaches used include some sort of wearable devices, ambient based devices or non-invasive vision based devices using live cameras. Most of such systems are either based on wearable or ambient sensor which is very often rejected by users due to the high false alarm and difficulties in carrying them during their daily life activities. This paper proposes a fall detection system based on an algorithm using combination of machine learning and human activity measurements such as changes of human height and rate of change of the subject during any of the activity. Classification of human fall from other activities of daily life is accomplished using height, changes in velocity and acceleration of the subject extracted from the depth information. Finally position of the subject and SVM classification is used for fall confirmation. From the experimental results, the proposed system was able to achieve an average accuracy of 97.39% with sensitivity of 100% and specificity of 96.61%.
|INVITED TALK #6
Shaparas Daliman '
and Syed Ab Rahman Abu-Bakar ''
' Universiti Malaysia Kelantan,
'' Universiti Teknologi Malaysia,
Recognition of Oil Palm Tree Based on Worldview-2 Images Using Haar-Based Rectangular Windows
Oil palm as one of the top commodities in Malaysia can be monitored and managed in a more effective, efficient and low-cost maintenance by using satellite imagery. The current standard of monitoring the number of oil palm trees is either by actual counting through deploying human workers at the plantation itself or by manually counting trees from the given airborne images. Such an approach seems to be cost inefficient and labour intensive in addition to high probability in recognition error. Therefore, this study aims to develop a technique for oil palm tree recognition and build a classification strategy for segmentation of oil palm tree area along with an approach to propose a structured flow mechanism for automatically counting the number of young and matured oil palm trees. A database of young, matured and non-oil palm tree objects based on high resolution (2m) in WorldView-2 images have been collected. It intends to identify features for development of oil palm tree recognition model by implementing object recognition techniques of Haar-based rectangular windows. As a result, it is found that classification based on features obtained from Haar-based rectangular windows has achieved 92.73% overall accuracy with 98.58% and 98.13% sensitivity to young and matured oil palm tree classification, respectively. Thus, it can be concluded that oil palm tree features derived from Haar-based rectangular windows is the most suitable for oil palm tree recognition model based on WorldView-2 images. Automated oil palm tree counting has achieved up to 100% accuracy with the ability to distinguish the non-oil palm tree objects during the counting.
|INVITED TALK #7
Bakhtiar Al Jefry Abd Salam
and Gan Hong Seng
Universiti Kuala Lumpur, British Malaysian Institute, Gombak, Selangor
Development of Minimal Intervention Semi-Automated Knee Cartilage Segmentation
Knee osteoarthritis is a common musculoskeletal disease resulted from the biochemical breakdown of articular cartilage in the synovial joints. The degenerative joint disease, for instance, causes mobility constraints such as walking or climbing stairs to affected patients, especially among the older population. Magnetic resonance imaging has been identified as a potential biomarker to comprehend the progression of osteoarthritis. Given the high anatomical complexity exhibited by knee cartilage, manual and conventional semi-automated segmentation will require tremendous human intervention. A background seed generation model has been proposed to mitigate manual labelling but the use of fixed threshold has invited inaccurate seed distribution. In this work, we propose to use Wellner’s method to determine the average threshold level for each superpixel based on saliency value. It can dynamically set average local threshold value according to the neighbouring superpixels. Integral image is an instrument where it can be utilized when we have a function from the pixel to real numbers (such as superpixel) and we wish to compute the sum of this function over a rectangular region of image. By calculating the integral image for each superpixel, it makes the computation of the integral image to become easier for the next stage. We applied the sum of function of the upper left corner and lower right corner of the neighbourhood superpixels to compute the average threshold value. The process was repeated until an optimal threshold value was obtained. We defined the saliency value of superpixel below the average threshold value as background, and above the threshold value as knee cartilage. Eventually, Wellner’s method of adaptive threshold can produce stable average threshold value to segment object and background.
|Back to Top|
The building (see image at the top of this page) is indicated by APU New campus (on Google Maps) or marked by Lifestyle by Modestos (on the map at the APU website). It is located along Jalan Teknologi 5, Technology Park Malaysia.
(Note that the building marked as Asia Pacific University of Technology & Innovation on Google Maps is the old campus, now APIIT & APLC).
Auditorium 1 is on the main level (Level 3, where the front stairways lead up to). Just walk along the walkway until you reach the cafeteria (Lifestyle by Modestos) area. The auditorium is on the right, opposite the cafeteria.
Parking: Limited visitors carparks are available at the campus. Although attempts will be made to secure free parking, charges may apply. Other open parking is available nearby between APU and Astro at the rate of RM 5.30 per entry.
|Back to Top|
|Back to Top|
|© 2017 IEEE Signal Processing Society Malaysia Chapter. All rights reserved.