Optical Remote Sensing Image Runway Extraction And Tank Target Recognition Technology

As a sort of important spatial information resource, the remote sensing images are widely used for resource exploration, military spying, inspection of environmental disaster, utilization of land, output assessment of crops and city planning etc, because of its effectiveness and practicability. It profoundly affects the national economy, national defense and development of the society.The optical remote sensing image is characterized by its large numbers of gray level, enormous amount information it delivers, the blurriness of boundaries, complex structure of the target, variable region-feature in condition of shooting environment. Because of these characters, there are no reliable models to guild the subtraction and the recognition of the targets in the optical remote sensing image. On the basis of the project"region segmentation and damage information extraction in the remote sensing image", this thesis consults the related literatures and commercial software home and abroad, performs further researches on the extraction of the run-way, recognition of the oilcan in the high spatial resolution optical remote sensing image, related work and achievement are as follow:1. Studies the basic algorithms of the digital image processing, for example: morphological dilation, erode, top-hat and low-hat transform, Gaussian blur filter, Canny segmentation, Otsu segmentation, traction of the boundaries, mark of connected domain. This thesis uses these mentioned algorithms in the preprocessing of the images and feature extraction of the targets, eliminates the disturbing features that the background brings, enhances the features of the desired targets. These algorithms are the basis of the extraction and recognition.2. Studies the run-way extraction technology. Run-way usually possess complex topology structure; have distinct boundaries with the surrounding area. This thesis studies the run-way extraction technology based on Hough transform line detection, and extract run-ways in the airport. The run-way extraction technology based on Hough transform line detection can extract the linear run-ways, but can't extract non-linear run-ways. On the basis of previous research, this thesis proposed a run-way extraction technology based on level set algorithm. Level set algorithm has the advantages of handling the complex topology structure, so the proposed technology not only can extract the linear run-ways, but also can extract the non-linear run-ways.3. Studies the oilcan recognition technology. This thesis studies oilcan recognition technology based on morphology. The oilcan recognition technology based on morphology can recognizes the oilcan when the desired targets are distinctive to the background, and with the prior information about the spatial location of the desired targets, the technology can even yields desirable result in the recognition of oilcans located on the complex background, but the technology is dependent on the prior information, and can't recognizes the desired targets that aren't distinctive to the background. On the basis of previous research, this thesis proposed a amended oilcan recognition technology based on Hough transform circle detection, the proposed technology makes use of the circular feature and non-overlap property of the oilcan targets, can recognizes the oilcans that are not distinctive to the background and have small shadow area when the edge of the targets can be detected by Canny operator.4. Finishes the software design and the code compilation of run-way extraction and oilcan recognition. The accuracy rates of the extraction and recognition are above 85%, time spent of the extraction or recognition is less than 2 minutes.