We are using the Autonomous Underwater Vehicle for imagery. The image rates is 4 images per second and the image resolutions is 2.35 megapixel.
Imaging is done line by line (more than 4000 images per line) and there is no any cross overlap between lines. Therefore, we need to have orthomosaic for each line.
The problem is when Pix4D run for 300 images (for instance), the result of the initial processing is like a spiral shape. Because of that the orthomosaic is not correct. But for a less group of images like 75 or 100 images, the output is more logic.
- Why Pix4D cannot detect the real position of images?
- Also, the photo direction is upward instead of downward as if the camera mounted above the vehicle, but actually, it looks downward to the bottom of sea.
I must mention that the vehicle records the position in each second. Therefore, all 4 images in same second has the same coordinate.
Also, I have read the other similar topics to find the solution and I tested all possibilities for calibration and matching processes, but finally I found that by decreasing the number of images the outputs will be more better. However, the photos direction do not change and are always up.
Could you please help me in this regard.