High-robustness sea-sky-line extraction method based on photoelectric sensor

Abstract

The invention relates to a high-robustness sea-sky-line extraction method based on a photoelectric sensor. The method comprises the following steps of: enhancing an image edge under a vertical gradient, filtering some irrelevant vertical line segments, sampling straight lines to obtain candidate points of a boundary, keeping direction information of the straight lines, extracting image blocks around the candidate points according to the candidate points of the boundary, and extracting the image blocks; and judging whether the boundary is the sea-sky-line or the edge of other objects according to the image feature information around the boundary point. A convolutional neural network is trained by using an image block containing a sea-sky-line and an image block only containing background information, so that the network can learn sea-sky-line features, the sea-sky-line image block and the background information image block are effectively distinguished in the operation of filtering boundary candidate points, and the image block is subjected to image segmentation in RANSAC straight line fitting. The Euclidean distance between the fitting straight line and the fitting point and direction information reserved by sampling the straight line direction and the fitting point from the straight line are considered, so that the sea-sky-line which is finally fitted is higher in precision and higher in robustness.

Description

Technical field [0001] The present invention relates to a photoelectric sensor and the unmanned boat automatic driving technology, and more particularly, to a method of extracting a high robustness sea antenna based photosensor. Background technique [0002] CN201710878146.X Chinese patent discloses a method for detecting antenna sea semantic segmentation based image, includes the following steps: (1) the input image to be detected; (2) using the algorithm SLIC superpixel dividing input image; (3) super pixels as a basic unit, probabilistic model of FIG sea image semantic segmentation, model and using the image from top to bottom in FIG sky is divided into regions, and the haze of the land area and the mixed area seawater; (4) sea water area extract to obtain a mask image seawater region; (5) on the region mask image seawater, seawater was extracted breakpoint region in the column direction based on the gradient information; (6) using the RANSAC algorithm to perform breakpoint re...

AI Technical Summary

Problems solved by technology

This method can accurately detect sea antennas in complex environments, and can also effectively detect coastlines in the background of long coasts. However, in some scenes where the boundary between sea and sky areas is not obvious, it is easily affected by noise such as boats, river banks, etc. Interference, for some objects with long straight lines such as river banks and boats, they will be considered as part of the sea antenna, which will affect the extraction accuracy

Images