Showing papers by "Heng-Da Cheng published in 1996"

Novel fuzzy logic approach to pavement distress detection

Abstract: In this paper, a novel pavement distress detection algorithm based on fuzzy logic is proposed. The idea of the proposed method is based on the fact that the crack pixels in pavement images are 'darker than their surroundings and continuous'. First, the proposed method determines how much darker the pixels are than the surroundings. This is done by determining the brightness membership function for gray levels in the difference image. Then, we check the connectivity of he darker pixels to eliminate the pixels which lack of connectivity. Finally, image projections are employed to classify cracks. The experimental results have shown that the cracks are correctly and effectively detected by the proposed method. The main advantages of the proposed method are: 1) It can correctly find out thin cracks even from very noisy pavement images. 2) It can be operated automatically. 3) The efficiency and accuracy of the proposed algorithm are superior. 4) Application-dependent nature, instead of image-dependent, will simplify the design of the system.© (1996) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Synthesized Images for Pavement Management System Design

Abstract: In visual pattern recognition, researchers often rely on standard test images obtained under fixed conditions because it is usually difficult to obtain images with all the desired characteristics. This practice can compromise an investigation because existing images may need major changes in optical parameters or the original scene may be unavailable. For most engineering applications, algorithms can only work well under certain conditions. New algorithms might require specifications not fully met by the existing data (patterns). Synthesized patterns, in which a computer fills these requirements, may be the answer to the problem, due to its advantages of low cost, easy specification, independence from environmental conditions, and realistic visualization of abstract ideas that stimulates creativity. Ray tracing--a computer graphics technique with optical correctness requiring a few numerical parameters--can be employed to devise a scene. Results are presented in the form of light intensity maps, convenient for pattern recognition algorithms. Once the algorithms under study are confirmed with the synthesized images, conditions can be specified for acquiring real-world images, thus reducing costly trial-and-error methods. This paper proposes a novel approach to generate cracks and potholes using "fractals" with "texture mapping." The results illustrate the practicality of the proposed approach in the design and implementation of pavement management systems.

A new approach to microcalcification detection in digital mammograms

Abstract: Computer-aided mammography is an important and challenging task in automated diagnosis. It has great potential over the traditional x-ray mammography in terms of efficiency and accuracy. Microcalcifications are the earliest sign of breast carcinomas and their accurate detection is one of the key issues for breast cancer control. In this study, a novel approach to detecting microcalcifications in digital mammograms is presented. The essential idea of the proposed approach is to apply a fuzzified image of a mammogram to locate the regions of interest and to interact the fuzzified image with the original image to preserve the fidelity. The major advantage of the proposed method is its ability to detect microcalcifications in very dense breast mammograms. A series of clinical mammograms are employed to test the proposed algorithm. The experiments aptly show that the microcalcifications are not only accurately detected but their features are also well preserved.

Snow classification from SSM/I data over varied terrain using an artificial neural network classifier

Abstract: The brightness temperatures (Tbs) observed by the Special Sensor Microwave/Imager (SSM/I) radiometer are sensitive to the changes in land surface snow conditions. Previously developed SSM/I snow classification algorithms have limitations and do not work properly for terrain where forests overlay snow cover. In this study, the authors applied unsupervised cluster analysis to define 6 snow classes in Tb observations, assessing both sparseand medium-vegetated region classes. Typical SSM/I Tb signature, in terms of cluster means, of each snow class was determined by calculating the mean Tbs of the corresponding cluster. A single-hidden-layer backpropagation (backprop) artificial neural network (ANN) classifier was designed to learn the 6 Tb patterns. Classification performance, in terms of error rate (%), was as small as 2.4%. This study confirms the potential of using cluster means in ANN supervised learning, and suggests a nonlinear retrieval method towards making the inferences of snow classes from SSM/I data over varied terrain operational. Improvement is expected by identifying more SSM/I Tb signatures of different land surface types to train the ANN classifier.