ing at the insufficient attention to influence of plant chlorophyll fluorescence on the accuracy of CO₂ inversion, we investigate the global vegetation fluorescence distribution, and simulate the influence of fluorescence column-averaged CO₂ dry-air mixing ratio (XCO₂). The simulation shows that when chlorophyll fluorescence is neglected, the inversion maximum error of XCO₂ can reach 15×10 ^-6. This bias can be controlled within 0.5×10 ^-6 by synchronous inversion fluorescence in a full-physics based retrieval. We retrieve the summer data of the Greenhouse Gases Observing Satellite (GOSAT) near Park Falls TCCON (The Total Carbon Column Observing Network) site. It is found that the error is corrected from 6×10 ^-6 to less than 2×10 ^-6 based on synchronous inversion fluorescence. This research shows that chlorophyll fluorescence cannot be neglected in high precision CO₂ inversion.

ing at the process of the registration of hyperspectral images and high spatial resolution images, it is difficult to choose the high-precision registration band because of the large difference between the bands of hyperspectral images. An algorithm for selecting high precision matching band of hyperspectral image based on Cram'er-Rao lower limit (CRLB) theory is proposed. Several bands with large amount of information and a small correlation in the hyperspectral image are selected by the band selection method. These bands are registered with the high spatial resolution image, respectively. The CRLB for each band's registration result is calculated. The high accuracy registration band is selected according to CRLB. The CRLB's registration quality evaluation performance is verified to be better by comparing CRLB and root mean square errors after each registration. And compared with the selected band registration results of CRLB and other methods, it is proved that the accuracy of the band registration selected by the proposed algorithm is high. The above band provides better data for the registration of hyperspectral images and high spatial resolution images.

ing at depth test of pipeline inwall defects, we propose a method of measuring the depth of pipeline defects, which based on the active thermal excitation by eddy current and infrared thermography. The theory of infrared imaging pipeline defects measuring is described. According to the special requirements of buried pipeline detection, a test device of eddy current thermal excitation with adjustable parameters is designed. Some specimens are fabricated according to the shape of the pipeline. With the active thermal excitation experiment based on eddy current, the influences of three important parameters, such as resonant frequency, lift-off distance and input electrical power on thermal excitation efficiency are analyzed, and the optimized values are obtained. Based on the above work, the infrared images of specimens with pre-designed defects which have different depths are acquired. The thermal image data analysis shows that the difference of grayscale between the defects and the non-defective areas varies with the defect depth, and the two factors show a single value correspondence, which has a good linearity under certain conditions. The defect depth detection model of groove-like and circular defects are established by the law. The experimental results show that the established model has certain detection accuracy. The research results show that the depth of defect can be calculated by infrared thermal image under the optimized active eddy current excitation condition. The proposed method based on active eddy current excitation of infrared thermal imaging pipeline is feasible.

ing at the occlusion phenomenon in the visual object, we propose a novel occlusion boundary detection approach for deep images based on the spectral clustering. Firstly, a new occlusion-related feature, effective standard deviation feature, is defined. Secondly, some pixels are extracted by using mean chi-square set distance, and the similarity matrix is constructed based on the occlusion-related feature. Thirdly, the Laplacian matrix of all the pixels and approximation eigenvectors are approximated by Nystrom approximation method based on the similarity matrix. Then, the obtained approximation eigenvectors are clustered to divide all the pixels in the depth image into two categories, namely the occlusion boundary points and non-occlusion boundary points. Finally, the occlusion boundary of the depth image is obtained by visualizing occlusion boundary points. Experimental results show that the proposed method which does not need any labeled samples has good effectiveness and generality for occlusion boundary detection of the object in the depth image.