Shyang Lih Chang

Video stabilization for a camcorder mounted on a moving vehicle

Vision systems play an important role in many intelligent transportation systems (ITS) applications, such as traffic monitoring, traffic law reinforcement, driver assistance, and automatic vehicle guidance. These systems installed in either outdoor environments or vehicles have often suffered from image instability. In this paper, a video stabilization technique for a camcorder mounted on a moving vehicle is presented. The proposed approach takes full advantage of the a priori information of traffic images, significantly reducing the computational and time complexities. There are four major steps involved in the proposed approach: global feature extraction, camcorder motion estimation, motion taxonomy, and image compensation. We begin with extracting the global features of lane lines and the road vanishing point from the input image. The extracted features are then combined with those detected in previous images to compute the camcorder motion corresponding to the current input image. The computed motion consists of both expected and unexpected components. They are discriminated and the expected motion component is further smoothed. The resulting motion is next integrated with a predicted motion, which is extrapolated from the previous desired camcorder motions, leading to the desired camcorder motion associated with the input image under consideration. The current input image is finally stabilized based on the computed desired camcorder motion using an image transformation technique. A series of experiments with both real and synthetic data have been conducted. The experimental results have revealed the effectiveness of the proposed technique.

A non-parametric blur measure based on edge analysis for image processing applications

A nonparametric image blur measure is presented. The measure is based on edge analysis and is suitable for various image processing applications. The proposed measure for any edge point is obtained by combining the standard deviation of the edge gradient magnitude profile and the value of the edge gradient magnitude using a weighted average. The standard deviation describes the width of the edge, and its edge gradient magnitude is also included to make the blur measure more reliable. Moreover, the value of the weight is related to image contrast and can be calculated directly from the image. Experiments on natural scenes indicate that the proposed technique can effectively describe the blurriness of images in image processing applications.

Nighttime pedestrian detection using thermal imaging based on HOG feature

This research focuses on pedestrian detection using infrared thermal imager. The purpose is to locate the pedestrians from studying thermal imagery. Based on HOG (Histograms of Oriented Gradients), Adaboost algorithm is used as a way to perform the detection. The system is divided into three sections, to extract the features of the pedestrians, to train the Adaboost classifier, and to detect the pedestrian. To get the features of the pedestrians, data is gathered from inserted images. The features allow the detection to work well. The feature extraction includes image segmentation, ROI selection, and feature extraction. We have successfully located the positions of the pedestrians with the methods mentioned above. This can be applied to the development of the intelligent driver assistance system, giving more road traffic situations to the drivers throughout the night.

Physics-based extraction of intrinsic images from a single image

A technique for extracting intrinsic images, including the reflectance and illumination images, from a single color image is presented. The technique first convolves the input image with a prescribed set of derivative filters. The pixels of filtered images are then classified into reflectance-related or illumination-related based on a set of chromatic characteristics of pixels calculated from the input image. Chromatic characteristics of pixels are defined by a photometric reflectance model based on the Kubelka-Munk color theory. From the classification results of the filtered images, the intrinsic images of the input image can be computed. Real images have been utilized in our experiments. The results have indicated that the proposed technique can effectively extract the intrinsic images from a single image.

Dichromatic reflection separation from a single image

A feature-based technique for separating specular and diffuse components of a single image is presented. In the proposed approach, Shafers dichromatic reflection model is utilized, which assumed a light reflected from a surface point is linearly composed of diffuse and specular reflections. The major idea behind the proposed method is to classify the boundary pixels of the input image to be specular-related or diffuse-related. A fuzzy integral process is proposed to classify boundary pixels based on their local evidences, including specular and diffuse estimation information. Based on the classification result of boundary pixels, an integration method is evoked to reconstruct the specular and diffuse components of the input image, respectively. Unlike previous researches, the proposed method has no color segmentation and iterative operations. The experimental results have demonstrated that the proposed method can perform dichromatic reflectance separation effectively with small misadjustments and rapid convergence.

Interference reflection separation from a single image

The interference image is defined as the superpositioning of a reflection image and an object image. A technique for separating reflection and object components of a single interference image is presented. The proposed method classifies edges of the interference image into either reflection or object related. Our method utilizes total variation (TV) method, blur measure, and region segmentation as evidence with a fuzzy integral technique to classify the edge pixels. Based on the results of edge pixel classification, the reflection and object components of the input image are reconstructed. Compared to previous published research, the proposed method is fast and requires no manual operations. The experimental results have demonstrated that the proposed method can perform separation of a single interference image effectively with small misadjustments and rapid convergence.

The invariance properties of chromatic characteristics

An approach to analyzing the degrees of invariance of chromatic characteristics is proposed in this paper. In many vision applications, it is desirable that the chromatic characteristics of objects in images taken under different lighting conditions could remain constant. However, the invariance properties of chromatic characteristics are subject to the lighting conditions. In order to be able to apply to dynamic scenes, we consider three fundamental lighting sources: diffuse, ambient, and directed lightings. Any illumination condition can be approximated as a combination of the three lighting sources. The proposed degree of chromatic invariance is defined based on the chromatic characteristic behaviors under different illumination conditions. A lot of image samples under different illumination conditions are utilized, and from experimental results, we conclude that chromatic characteristics {H, C, Cλ} are most stable and suitable for the vision applications.

Machine-vision based obstacle avoidance system for robot system

In this paper, we proposed a machine-vision based obstacle avoidance system for robot system by using single camera, it could accomplished an obstacle avoidance and path planning. The structure of this system is using a camera and two laser projectors fixed on same base. When robot get into a unknown environment, it will stop and capture an image, the system use several simple image process steps to recognize the obstacle. The system will rotate the base to project the laser points on obstacle, the distance measurement results calculated by IBDMS method, when the system calculated the distance between obstacle and robot, it could planned the path to achieve the autonomous patrol.

A Data Transfer Integrate System by Using Single Camera

there is a novel data transfer integration system by using single image proposed in this paper. There are many sensors use to monitoring the landslide occurred or not, but the monitoring results have different transmit protocol such as: Wi-Fi, transmit line, 2.4 GHz and ZigBee. This system is a data transfer integration system for all sensors, and it can achieve a real time monitoring system. The system integrate many sensor signals to microprocessor, and convert analog signal to digital signal by using A/D converter, i2c and UART, and the measuring results of sensors and display on a 8*8 matrix LED board. The system use the existing landslide-monitoring camera, and the matrix LED board install in the monitoring area. The result of sensors can be transmit to monitoring center by using single image, the data will not affect by environment, weather and mountain terrains.

Dangerous Prediction by Case-Based Approach on Expressways

The purpose of this system is to reduce the rate of dangerous events caused by various driving factors. We compose a driving relational map as the system inputs factors of driver behavior, nearby vehicles and roadway factors, and then put this driving relational map into the matching process with dangerous driving relational maps. If the similarity between the driving relational map and one of the dangerous driving relational maps is high, a dangerous event may occur. At this time the system will warn the driver to watch out for this dangerous event. Along with the learning process based on case-base reasoning, the system will become a flawless danger prediction system.