Pakorn Kaewtrakulpong

An Improved Adaptive Background Mixture Model for Real- time Tracking with Shadow Detection

Real-time segmentation of moving regions in image sequences is a fundamental step in many vision systems including automated visual surveillance, human-machine interface, and very low-bandwidth telecommunications. A typical method is background subtraction. Many background models have been introduced to deal with different problems. One of the successful solutions to these problems is to use a multi-colour background model per pixel proposed by Grimson et al [1, 2,3]. However, the method suffers from slow learning at the beginning, especially in busy environments. In addition, it can not distinguish between moving shadows and moving objects. This paper presents a method which improves this adaptive background mixture model. By reinvestigating the update equations, we utilise different equations at different phases. This allows our system learn faster and more accurately as well as adapts effectively to changing environment. A shadow detection scheme is also introduced in this paper. It is based on a computational colour space that makes use of our background model. A comparison has been made between the two algorithms. The results show the speed of learning and the accuracy of the model using our update algorithm over the Grimson et al’s tracker. When incorporate with the shadow detection, our method results in far better segmentation than The Thirteenth Conference on Uncertainty in Artificial Intelligence that of Grimson et al.

A real time adaptive visual surveillance system for tracking low-resolution colour targets in dynamically changing scenes

Abstract This paper presents a variety of probabilistic models for tracking small-area targets which are common objects of interest in outdoor visual surveillance scenes. We address the problem of using appearance and motion models in classifying and tracking objects when detailed information of the objects appearance is not available. The approach relies upon motion, shape cues and colour information to help in associating objects temporally within a video stream. Unlike previous applications of colour and complex shape in object tracking, where relatively large-size targets are tracked, our method is designed to track small colour targets commonly found in outdoor visual surveillance. Our approach uses a robust background model based around online Expectation Maximisation to segment moving objects with very low false detection rates. The system also incorporates a shadow detection algorithm which helps alleviate standard environmental problems associated with such approaches. A colour transformation derived from anthropological studies to model colour distributions of low-resolution targets is used along with a probabilistic method of combining colour and motion information. A data association algorithm is applied to maintain tracking of multiple objects under circumstances. Simple shape information is employed to detect subtle interactions such as occlusion and camouflage. A novel guided search algorithm is then introduced to facilitate tracking of multiple objects during these events. This provides a robust visual tracking system which is capable of performing accurately and consistently within a real world visual surveillance arena. This paper shows the system successfully tracking multiple people moving independently and the ability of the approach to maintain trajectories in the presence of occlusions and background clutter.

Jeremiah: the face of computer vision

This paper presents a humanoid computer interface (Jeremiah) that is capable of extracting moving objects from a video stream and responding by directing the gaze of an animated head toward it. It further responds through change of expression reflecting the emotional state of the system as a response to stimuli. As such, the system exhibits similar behavior to a child. The system was originally designed as a robust visual tracking system capable of performing accurately and consistently within a real world visual surveillance arena. As such, it provides a system capable of operating reliably in any environment both indoor and outdoor. Originally designed as a public interface to promote computer vision and the public understanding of science (exhibited in British Science Museum), Jeremiah provides the first step to a new form of intuitive computer interface.

Adaptive Visual System for Tracking Low Resolution Colour Targets.

This paper addresses the problem of using appearance and motion models in classifying and tracking objects when detailed information of the object’s appearance is not available. The approach relies upon motion, shape cues and colour information to help in associating objects temporally within a video stream. Unlike previous applications of colour in object tracking, where relatively large-size targets are tracked, our method is designed to track small colour targets. Our approach uses a robust background model based around Expectation Maximisation to segment moving objects with very low false detection rates. The system also incorporates a shadow detection algorithm which helps alleviate standard environmental problems associated with such approaches. A colour transformation derived from anthropological studies to model colour distributions of low-resolution targets is used along with a probabilistic method of combining colour and motion information. This provides a robust visual tracking system which is capable of performing accurately and consistently within a real world visual surveillance arena. This paper shows the system successfully tracking multiple people moving independently and the ability of the approach to recover lost tracks due to occlusions and background clutter.

Object detection and modeling algorithm for automatic visual people counting system

This paper presents object detection and modeling algorithm for automatic visual people counting system to identify individuals from top view images acquired from an overhead surveillance camera. This work proposes Snake algorithm with effective external energy function and a half-circle template initialization for modeling passengers. The experiment result showed that the proposed method could fit the Snake to the head and shoulder boundaries of the passengers effectively. Fitting error of 3.76 pixels per control point was obtained in our experiment. The information is not only useful for verifying humans but also for extracting appearance features to enhance performance of the subsequent tracking algorithm.

A Real-Time Automatic Inspection System for Pattavia Pineapples

This paper presents a machine vision method to inspect the maturity of pineapples that ripe naturally. Unlike previous methods, the proposed technique can be categorized as a real-time non destructive testing (Real-Time NDT) approach. It consists of two phases, learning and recognition phases. In the learning phase, the system constructs a library of reference pineappleskin- color models. In the recognition phase, the same process is performed to build a pineappleskin- color model of the testing subject. The model is then compared with each of the reference in the library by a method called region-segmented histogram intersection. The subject is then labeled with the grade of the best match. The system achieved a high performance and speed (3 frames/sec.) in our experiment. The system also includes weighing machine on belt transmission for weight prediction.

Semi-Parametric Learning for Classification of Pitting Corrosion Detected by Acoustic Emission

This paper presents a Non-Destructive Testing (NDT) technique, Acoustic Emission (AE) to classify pitting corrosion severity in austenitic stainless steel 304 (SS304). The corrosion severity is graded roughly into five levels based on the depth of corrosion. A number of timedomain AE parameters were extracted and used as features in our classification methods. In this work, we present practical classification techniques based on Bayesian Statistical Decision Theory, namely Maximum A Posteriori (MAP) and Maximum Likelihood (ML) classifiers. Mixture of Gaussian distributions is used as the class-conditional probability density function for the classifiers. The mixture model has several appealing attributes such as the ability to model any probability density function (pdf) with any precision and the efficiency of parameter-estimation algorithm. However, the model still suffers from model-order-selection and initialization problems which greatly limit its applications. In this work, we introduced a semi-parametric scheme for learning the mixture model which can solve the mentioned difficulties. The method was compared with conventional Feed-Forward Neural Network (FFNN) and Probabilistic Neural Network (PNN) to evaluate its performance. We found that our proposed methods gave much lower classificationerror rate and also far smaller variance of the classifiers.

Corrosion-source location by an FPGA-PC based acoustic emission system

This paper presents a novel low-cost acoustic emission (AE) source-location system to locate corrosion sources in AISI304 austenitic stainless steel. The system is implemented using an FPGA-PC configuration. Three AE sensors with 150 kHz resonance frequency were used to detect the AE activities generated from the corrosion. Experiments were set up to show performance of the system in locating uniform and pitting corrosions in stable state using corrosive solutions and electrochemical environment. A commercial AE system was used to assure the AE activities with the location results. Experimental results showed the ability of the AE source location system to locate corrosion sources. In conclusion, the proposed AE source-location system is appropriate and practicable to locate corrosion with flexibility and affordability.

An efficient translation-rotation template matching using pre-computed scores of rotated templates

Printed Circuit Board Assembly (PCBA) quality inspection is an essential stage in PCBA manufacturing. In this paper, we focus on development of efficient template matching algorithms used in Automated Optical Inspection (AOI) for PCBA quality inspection in post-reflow process. In our work, a pre-computed set of Normalized Cross Correlation (NCC) scores from rotated templates to the original template are used to eliminate unnecessary calculation and to estimate rotation angles of scene object images. Two models called multiple band and piecewise linear are implemented and compared to find suitable rotation angles of candidate locations. Since the technique follows traditional systematic window sliding, existing efficient implementation techniques of template matching can be directly applied. Unlike other rotation invariant methods, accurate rotation angles can be directly obtained from the technique. Experimental results show excellent performance for PCBA quality inspection applications.

Fitting a Pineapple Model for Automatic Maturity Grading

In this paper, we present a pineapple skin model and a method to fit the model. Our main application of the model is for automatic maturity grading of pineapples in canned pineapple industry. The model consists of two subparts: Phyllotaxis and pineapple scale models. The Phyllotaxis model represents the spiral arrangement of pineapple-scales, which is a growing pattern of the fruit. It includes a string of scale-model cells. The scale model includes boundary, internal area and petal part of scale. Modified snake algorithm is used to construct the structure model while Active Shape Model (ASM) is applied to each scale. The model can accurately fit to pineapple skins in our experiment and classification features of the fruit can be extracted.