Sang-Woo Ban

2008 Special Issue: Stereo saliency map considering affective factors and selective motion analysis in a dynamic environment

We propose new integrated saliency map and selective motion analysis models partly inspired by a biological visual attention mechanism. The proposed models consider not only binocular stereopsis to identify a final attention area so that the system focuses on the closer area as in human binocular vision, based on the single eye alignment hypothesis, but also both the static and dynamic features of an input scene. Moreover, the proposed saliency map model includes an affective computing process that skips an unwanted area and pays attention to a desired area, which reflects the human preference and refusal in subsequent visual search processes. In addition, we show the effectiveness of considering the symmetry feature determined by a neural network and an independent component analysis (ICA) filter which are helpful to construct an object preferable attention model. Also, we propose a selective motion analysis model by integrating the proposed saliency map with a neural network for motion analysis. The neural network for motion analysis responds selectively to rotation, expansion, contraction and planar motion of the optical flow in a selected area. Experiments show that the proposed model can generate plausible scan paths and selective motion analysis results for natural input scenes.

Explosive gas recognition system using thick film sensor array and neural network

Abstract A sensor array with nine discrete sensors integrated on a substrate was developed for recognizing the species and quantity of explosive gases such as methane, propane, and butane. The sensor array consisted of nine oxide semiconductor gas-sensing materials with SnO 2 as the base material plus a heating element based on a meandered platinum layer all deposited on the sensor. The sensors on the sensor array were designed to produce a uniform thermal distribution and show a high and broad sensitivity and reproductivity to low concentrations through the use of nano-sized sensing materials with high surface areas and different additives. Using the sensitivity signals of the array along with an artificial neural network, a gas recognition system was then implemented for the classification and identification of explosive gases. The characteristics of the multi-dimensional sensor signals obtained from the nine sensors were analyzed using the principal component analysis (PCA) technique, and a gas pattern recognizer was implemented using a multi-layer neural network with an error back propagation learning algorithm. The simulation and experimental results demonstrate that the proposed gas recognition system is effective in identifying explosive gases. For real time processing, a DSP board (TMS320C31) was then used to implement the proposed gas recognition system in conjunction with a neural network.

SnO/sub 2/ gas sensing array for combustible and explosive gas leakage recognition

A gas-sensing array with ten different SnO/sub 2/ sensors was fabricated on a substrate for the purpose of recognizing various kinds and quantities of indoor combustible gas leakages, such as methane, propane, butane, LPG, and carbon monoxide, within their respective threshold limit value (TLV) and lower explosion limit (LEL) range. Nano-sized sensing materials with high surface areas were prepared by coprecipitating SnCl/sub 4/ with Ca and Pt, while the sensing patterns of the SnO/sub 2/-based sensors were differentiated by utilizing different additives. The sensors in the sensor array were designed to produce a uniform thermal distribution along with a high and differentiated sensitivity and reproducibility for low concentrations below 100 ppm. Using the sensing signals of the array, an electronic nose system was then applied to classify and identify simple/mixed explosive gas leakages. A gas pattern recognizer was implemented using a neuro-fuzzy network and multi-layer neural network, including an error-back-propagation learning algorithm. Simulation and experimental results confirmed that the proposed gas recognition system was effective in identifying explosive and hazardous gas leakages. The electronic nose in conjunction with a neuro-fuzzy network was also implemented using a digital signal processor (DSP).

Dynamic visual selective attention model

We propose a new biologically motivated dynamic bottom-up selective attention model, which can generate a saliency map (SM) by considering dynamics of continuous input scenes as well as saliency of the primitive features of a static input scene. The maximum entropy algorithm is used to develop the dynamic selective attention model, whereby the input consists of the static bottom-up SMs for the successive static scenes. The experimental results show that the proposed model can generate more plausible scan paths for a dynamic scene compared with those obtained by the static bottom-up attention model.

Micro gas sensor array with neural network for recognizing combustible leakage gases

A micro gas sensor array, consisting of four porous tin oxide thin films added with noble metal catalysts on a micro-hotplate, was designed and fabricated. The micro-hotplate was designed to obtain a uniform thermal distribution along with a low-power consumption and fast thermal response. The sensing properties of the sensors toward certain combustible gases, i.e., propane, butane, LPG, and carbon monoxide, were evaluated. A multilayer neural network was then used to classify the gas species. The results demonstrated that the proposed micro sensor array, plus multilayer neural network employing a backpropagation learning algorithm, was very effective in recognizing specific kinds and concentration levels of combustible gas below their respective threshold limit values.

Improving AdaBoost Based Face Detection Using Face-Color Preferable Selective Attention

In this paper, we propose a new face detection model, which is developed by combining the conventional AdaBoost algorithm for human face detection with a biologically motivated face-color preferable selective attention. The biologically motivated face-color preferable selective attention model localizes face candidate regions in a natural scene, and then the Adaboost based face detection process only works for those localized face candidate areas to check whether the areas contain a human face. The proposed model not only improves the face detection performance by avoiding miss-localization of faces induced by complex background such as face-like non-face area, but can enhances a face detection speed by reducing region of interests through the face-color preferable selective attention model. The experimental results show that the proposed model shows plausible performance for localizing faces in real time.

Affective saliency map considering psychological distance

This paper proposes a new affective saliency map (SM) model considering psychological distance as well as the pop-out property based on relative spatial distribution of the primitive visual features such as intensity, edge, color, and orientation. By reflecting congruency between the spatial distance caused by spatial proximity and distal in a visual scene and psychological distance caused by the way people think about visual stimuli, the proposed SM model can produce more human-like visual selective attention than a conventional SM model based on primary visual perception. In the proposed model, a psychological distance caused by a social distance, in which a proximal entity such as friend becomes more attractive when it is located near but a distal entity such as enemy becomes more attractive when it is located far from an observer, is considered. In the experiments, two types of visual stimuli are considered, mono-stimuli and stereo-stimuli. In the case of mono-stimuli, the visual stimuli on a picture with psychological depth cues were considered. Instead, in the case of stereo-stimuli, depth perception is also considered for obtaining real spatial distance of visual target in a visual scene. In order to verify the proposed affective SM model, an eye tracking system was used to measure the visual scan path and fixation time on a specific local area while monitoring the visual scenes by human subjects. Experimental results show that the proposed model can generate plausible visual selective attention properly reflecting both psychological distance and primitive visual stimuli inducing pop-out bottom-up features.

A face detection using biologically motivated bottom-up saliency map model and top-down perception model

Abstract We propose a new biologically motivated face detection model. The proposed model integrates a bottom-up mechanism for extracting features to obtain salient information and a top-down perceptual mechanism for perceiving face features such as face form and face color. We constructed a face conspicuity map by binding the bottom-up process and the top-down process that consists of face form and color feature maps. Computer experimental results show that the proposed model successfully indicates faces in natural scenes.

BIOLOGICALLY MOTIVATED INCREMENTAL OBJECT PERCEPTION BASED ON SELECTIVE ATTENTION

In this paper, we propose an object selective attention and perception system, which was implemented by integrating a specific object preferable attention model with an incremental object perception model. The object oriented attention model can selectively pay attention to the candidates of an object in natural scenes based on a bottom-up selective attention model in conjunction with a top-down biased attention mechanism for a specific object. A generative model based on an incremental Bayesian parameter estimation is considered in order to perceive arbitrary objects in the attended areas. Combining an object oriented attention model with general object perception model, the developed system cannot only pay attention to a specific target object but can also memorize the characteristics of task nonspecific objects in an incremental manner. Experimental results show that the developed system generates good performance in successfully focusing on the target objects as well as incrementally perceiving objects in natural scenes.

Real time implementation of a selective attention model for the intelligent robot with autonomous mental development

We propose a biologically motivated selective attention model to find an object based on context free search for an intelligent robot with an autonomous mental development (AMD) mechanism. For real-time operation of the selective attention model in the robot system, we have considered a way to reduce the computational load of the selective attention model, which uses a simplified symmetry operation with retina-topic sampling and look-up table in the localized candidate attention region. As a result, our model can perform within 270 ms at Pentium-4 2.8Ghz, and obtain a plausible human-like visual scan path in order to pay attention to an object preferentially. Then, we implemented an intelligent mobile robot with selective attention for an AMD mechanism.