Pcnn Is, And Zernike Moments In Remote Sensing Image Processing And Target Recognition Applications

Now, computer vision based on pattern recognition is most active area of digital image processing research and application, while target recognition is an important research direction in computer vision, and it's widely applied in many fields, such as industrial automation, military sensing. Aircraft and vessel recognition in sensing image has great sense in military sensing. In this paper, we studied Pulse Coupled Neural Network's (PCNN) model, theory in image segmentation, and existed defects. Based on other researches, we improved PCNN model for its application in remote sensing image segmentation. Besides, we analyzed Hu moment and Zernike moment's theory, character and application in pattern recognition. We also improved Zernike moment, let it had two more characters of scaling invariance and translating invariance, while original moment only has character of rotating invariance.At last, we studied labeling algorithm of binary image, and researched plane recognition algorithm. In this algorithm, we use PCNN to segment remote sensing images; and then label the segmented result; finally, use Zernike moment to stand for targets'shape features and match with template for target identification. Results show that the algorithm has the properties of high recognizing rate and stability, and can against effects of translation, rotations, small scaling and shadow, etc. at the same time, we also proposed a vessel recognizing algorithm. Aiming at effects of shadows and linked vessels in remote sensing images, we proposed a solution.