Entropy-based Deep-sea Resources Image Processing Algorithm Research And Applications

Image processing is a critical technology for achieving the mining coverage in the deep-sea. It is the basis of moving the mining vehicle optimally and optimizing the process parameters. Researching on the deep-sea resource image processing is very important for improving the capacity of environment recognition about mine vehicle. This thesis is mainly concentrated on the processing algorithms of the deep-sea image, including image enhancement, image segmentation. And the coverage of mining is calculated by image processing. Using DSP (Digital Signal Processor), these algorithms can achieve the high speed running. And the algorithm procedure can be easily transplanted and upgraded.The main work is that the entropy information is widely used in the processing of deep-sea image. The method of fuzzy enhancement on deep-sea image based on relative entropy is proposed. The problem of defining the cross point is well dealt through using the combination of the theory of image segmentation and relative entropy. In order to improve efficiency of segmentation and achieve accurate coverage, a method combing 2D maximum entropy is introduced. Owning to the uneven brightness in the deep-sea, the whole image of mining could not be segmented efficiently, lastly the image block is used. The method combing 2D maximum entropy is used in these blocks of the whole image. The method expresses a good effect to images segmentation.In order to satisfy the real time request in the deep-sea mining, A DSP chip TMS320DM642 is used as core computer to design an image processing system. The paper completes the arithmetic transplant from PC to DSP and prepares well for application to practical systems.