Design And Application Of The Parallel Labeling Algorithm Of 8-adjacent Connecting Area For Massive Gray Scale Images

For remote sensing image data, such as the biophysical parameter product, the biological environment monitoring product in large project evaluation area, the plant dominant community monitoring in the National Nature Reserve, the early warning product of habitat fragmentation in the National Nature Reserve, and the remote sensing application of regional ecological environment disaster, etc, the work of data producting, processing and handling has two steps: First step is to carry out remote sensing image segmentation target data, and to extract the information they need through target identification. Second step is to analyze the splitting target results, and target recognition processing, a process known as connected component labeling. It is a important question that how to make communication between border regions exhibit smoothly, natural-looking results when rapid parallel communication domain. Therefore, in order to achieve the communication area boundary massive remote sensing image data showing a natural, smooth effect, optimize the original four-way communication to eight-way communication. However, in dealing with some of the large amount of data and computation of massive remote sensing data(gray scale image), the currently existing algorithms can not meet the demand. To solve this problem, a study massive remote sensing data for rapid, efficient parallel algorithms become the research emphasis.To solve the problems, this study makes it through massive parallel mark algorithm based on the using of remote sensing data and consider how to apply the algorithm in the actual development process of the "high-resolution Earth observation system". Thre are the main contents and contributions of my research work.(1) Proposed a parallel algorithm that marking the eight massive remote sensing data communication field. For there are connected component labeling algorithm in remote sensing data to quickly tag to eight connected domain basis NA Questions and existing connected component labeling algorithm, it will be the traditional eight-connected component labeling algorithm to optimize from the direction(eight to Connectivity optimized for four-way communication), the conflict in the process of communicating mark, the use of list processing mechanism to resolve conflict, while taking advantage of block processing and merge processing mechanism, not only to achieve massive remote sensing image data of the image correctly in communication domain partition, and can accurately and effectively calculate the number of connected domains, and massive remote sensing image data at massive remote sensing image data to communicating effectively eight domains marked speed parallel processing, resulting in a complete processed. By the experimental results show that the massive gray image based on 8-adjacent connecting area parallel labeling algorithm proposed in this paper can not only apply to the current processing requirements, and is highly efficient in dealing with massive remote sensing image data.(2)The actual application and development process in the major projects: "High-resolution Earth observation system". In the actual development and design process, in order to make "high-resolution Earth observation system" can effectively addressing the remote sensing image data segmentation target recognition and information extraction. We used the mass communication field parallel eight gray scale image labeling algorithm successfully in the rsearch and development work of the project. It works well in the processing and producting of remote sensing products such as the biophysical parameter product, the biological environment monitoring product in large project evaluation area, the plant dominant community monitoring in the National Nature Reserve, the early warning product of habitat fragmentation in the National Nature Reserve, and the remote sensing application of regional ecological environment disaster, etc.