Objective: At present, it is faced with the problem of large amountof high-precision remote sensing image basic data and weak real-time update.
In this paper, an intelligent generation method of mosaic line is proposed,
which takes into account the features of ground objects, After generating
mosaic lines, update mosaic based on existing large area base map is realized.
This method can meet the real-time updating requirements of high-precision
orthophoto. Method: The proposed method is based on the optimized model
of grabcut image segmentation proposed by Carsten rother et al. In this paper,
an improved optimization model of grabcut is proposed to generate mosaic
line segments, Then the segmented image is segmented, and the embedded line
segments are intelligently bypassed by buildings and other features through
topology checking. Through the combination of the generated four mosaic line
segments, a complete mosaic line is generated. Result: Different strategies are
used to generate mosaic lines for different areas of surface features. Experiments
show that the above method can generate mosaic lines efficiently, and the
performance of this method is better than the traditional mosaic line algorithm.
Conclusion In this paper, an intelligent generation strategy of mosaic line is
proposed, which takes into account the surface features. In view of the existing
regional base map, real-time single scene image is used to update the mosaic
line. Mosaic line has high timeliness. The experimental results show that the
timeliness is about 7.3 square kilometers per second when the remote sensing
image data with 1 meter resolution are dynamically updated and embedded,
and the updated mosaic results can meet the application requirements of highprecision orthophoto real-time dynamic update.