A Research On The Infrared Image's Collection And Segmentation Technology

Under the complicated circumstance, segmenting the infrared targets precisely and automatically is very difficult. For segmenting the infrared targets precisely and timely without losing the minutiae of target, man-machine interaction operation is always needed. On the background of 973-project, this thesis makes a deep research on the collection and segmentation of the infrared image. The main content of this thesis is as following:In Chapter 1 a basic devising project of the infrared image collection card is provided based on PCI bus according to the system requirements. Then the paper reviews the development and status quo of infrared image segmentation and analyses the significance of man-machine interaction in image segmentation according to the characters of target segmentation under complicated circumstance.In Chapter 2 we analyze the requirements of the infrared image collection card in detail, design and develop it concretely, which takes the generality of the system into account and adopts the combined method using the PCI local bus and large-scale programming logic device. The programming of the driver is expressed in the thesis and the communication between the collection card and application is carried out.Combining with the characteristics of the extended infrared target under the complicated circumstance, the technology of man-machine interaction into the system is performed in Chapter 3. Given a improved EFT algorithm based on the detailed analysis of the image foresting transform (LFT) algorithm. Firstly, a threshold was given by the Otsu algorithm. Secondly, modify the engaged qualification of the IFT path cost, given a modified algorithm that can reduce the memory space and advance the speed. The result shows the separation between the target and background, and the minutiae are more intact.The general work of this thesis is summarized and the farther direction is pointed out in the last chapter.