CN-115631403-B - Row overlap width detection method and system based on sonar image

CN115631403BCN 115631403 BCN115631403 BCN 115631403BCN-115631403-B

Abstract

The invention provides a row overlap width detection method and system based on a sonar image, wherein the method comprises the steps of obtaining a first sonar image containing a target row, inputting the first sonar image into a preset sonar image processing neural network model, determining a contour line and a turning plate rotating frame corresponding to the target row, wherein the long side of the turning plate is perpendicular to the laying direction of the target row, determining an overlap point and a boundary line of the target row according to the contour line corresponding to the target row, and determining the overlap width of the laying according to the overlap point and the boundary line. Because the turning plate has obvious characteristics in the sonar image, the long side of the turning plate is perpendicular to the laying direction of the target row, and the boundary line of the target row can be better determined by combining the image information of the turning plate, so that the identification precision of the sonar image is improved, and the generalization capability of row overlap width detection based on the sonar image is improved.

Inventors

LUO MINGYU
Yi Qiuchen
LIN JIAN

Assignees

东云睿连(武汉)计算技术有限公司

Dates

Publication Date: 20260508
Application Date: 20221019

Claims (8)

1. The row overlap width detection method based on the sonar image is characterized by comprising the following steps of: Acquiring a first sonar image containing a target row; Inputting the first sonar image into a preset sonar image processing neural network model, and determining an outline corresponding to the target row and a turning plate rotating frame corresponding to the target row according to the sonar image processing neural network model, wherein the long side of the turning plate is perpendicular to the laying direction of the target row; determining a lap point of the target row and a boundary line of the target row according to the contour line corresponding to the target row and the turning plate; Determining the overlapping width of the paving row according to the overlapping point and the boundary line; wherein, the determining the overlap point of the target row according to the contour line corresponding to the target row and the turning plate comprises: Determining a boundary between the target row and a riverbed area in the first sonar image according to a contour line corresponding to the target row; Determining an overlap point of the target row according to the dividing line, wherein the overlap point is positioned in the overlap area of the target row and the row adjacent to the target row; wherein, the determining the boundary line of the target row according to the contour line corresponding to the target row and the turning plate comprises the following steps: Determining a marker with a deterministic included angle with the laying direction of the target row in the first sonar image according to the contour line corresponding to the target row; Calculating a slope of the boundary line from the marker; Selecting a point which is close to the overlap area and is positioned on the boundary line as an anchor point; and determining the boundary line according to the slope of the boundary line and the anchor point.
2. The method of claim 1, wherein said determining a contour line corresponding to the target object from the sonar image processing neural network model comprises: inputting the first sonar image into a preset sonar image processing neural network model, and obtaining a second sonar image according to the sonar image processing neural network model, wherein the second sonar image comprises a contour line corresponding to the target bank.
3. The method according to claim 1, wherein a plurality of rows are sequentially connected and arranged on the river bed, the rows comprise a laid row and a row to be laid, and the boundary line of the target row is a boundary line formed between one side of the laid row, which is close to the current row to be laid, and the river bed.
4. The method of claim 1, wherein said determining the overlap width of the row from the overlap point and the boundary line comprises: calculating a first distance value from the overlapping point to the boundary line, wherein the first distance value is a pixel distance in the first sonar image; and converting the first distance value into a second distance value according to a preset scale, wherein the second distance value is the overlap joint width of the row body.
5. The method of claim 1, wherein training the pre-set sonar image processing neural network model comprises: acquiring training data, wherein the training data comprises a third sonar image containing a target bank; labeling the third sonar image, and determining the contour line of the target row and a turning plate rotating frame corresponding to the target row; Training the preset sonar image processing neural network model according to the third sonar image, the contour line of the target row and the turning plate rotating frame corresponding to the target row.
6. A sonar image-based row overlap width detection system for implementing the sonar image-based row overlap width detection method of any one of claims 1-5, comprising: The data acquisition assembly is used for acquiring a first sonar image containing a target row; The imaging analysis component is used for inputting the first sonar image into a preset sonar image processing neural network model, determining an outline corresponding to the target row and a turning plate rotating frame corresponding to the target row according to the sonar image processing neural network model, wherein the long side of the turning plate is perpendicular to the laying direction of the target row; and the drawing component is used for providing an interactive drawing interface and outputting the boundary line corresponding to the target row, the vertical line segment from the overlap point to the boundary line and the value of the overlap width in the first sonar image.
7. An electronic device comprising a processor, a memory and a computer program stored on the memory and capable of running on the processor, which when executed by the processor performs the steps of the method according to any of claims 1-5.
8. A computer readable storage medium, characterized in that it has stored thereon a computer program which, when executed by a processor, implements the steps of the method according to any of claims 1-5.

Description

Row overlap width detection method and system based on sonar image Technical Field The invention relates to the field of sonar underwater detection and identification, in particular to a row overlap width detection method and system based on a sonar image. Background The traditional row overlap width detection method mainly comprises the steps of shallow contact of a diver, visual observation or imaging by means of an underwater camera is used for observing whether a distance warning line which is placed in advance between adjacent rows is visible, if so, the overlap distance is too small, and the laying operation is not qualified. The manual underwater operation has the defects of high safety risk and high labor cost, is additionally influenced by water flow scouring sediment, and can be unclear in visual or underwater image, so that the detection result is inaccurate. The current mainstream method is that after imaging the working area of the water bottom by using sonar scanning while laying and arranging work, a detector judges and reads the image according to experience, marks the boundary line between the upper row body and the river bed and the lap joint line (point) between the two rows, draws a vertical line to the boundary line through the point, and converts the length of the vertical line section by a scale to be used as the lap joint width of actual detection. However, the existing method still has the conditions of insufficient precision and poor generalization in the detection of the overlapping width of the sonar image array. Conventional image morphology algorithms rely too much on morphological features, such as sharpness, shadows, illumination, etc., that are present only on part of the image, so that the morphological transformation parameters used only take effect in certain situations. Other general sonar image target detection algorithms aim at identifying a stationary or moving tangible object, and the row bodies are often paved on the whole sonar image in a piece-wise manner, individuals cannot be distinguished, and the lap joint area of the row bodies is not provided with accurate object characteristics. Disclosure of Invention In view of the foregoing, embodiments of the present invention are provided to provide a method for detecting a overlap width of a row based on a sonar image, and a corresponding apparatus for detecting a overlap width of a row based on a sonar image, which overcome or at least partially solve the foregoing problems. In order to solve the above problems, on the one hand, an embodiment of the invention discloses a row overlap width detection method based on sonar images, which comprises the following steps: Acquiring a first sonar image containing a target row; Inputting the first sonar image into a preset sonar image processing neural network model, and determining an outline corresponding to the target row and a turning plate rotating frame corresponding to the target row according to the sonar image processing neural network model, wherein the long side of the turning plate is perpendicular to the laying direction of the target row; determining a lap point of the target row and a boundary line of the target row according to the contour line corresponding to the target row and the turning plate; And determining the overlapping width of the pavement according to the overlapping point and the boundary line. Optionally, the determining, according to the sonar image processing neural network model, a contour line corresponding to the target object includes: inputting the first sonar image into a preset sonar image processing neural network model, and obtaining a second sonar image according to the sonar image processing neural network model, wherein the second sonar image comprises a contour line corresponding to the target bank. Optionally, the plurality of rows are connected in turn and are arranged on the riverbed, the rows comprise paved rows and rows to be paved, and the boundary line of the target row is a boundary line formed between one side of the paved row close to the current row to be paved and the riverbed. Optionally, the determining, according to the contour line corresponding to the target row and the turning plate, the overlap point of the target row includes: Determining a boundary between the target row and a riverbed area in the first sonar image according to a contour line corresponding to the target row; And determining the lap joint point of the target row body according to the dividing line, wherein the lap joint point is positioned in the lap joint area of the target row body and the row body adjacent to the target row body. Optionally, the determining the boundary line of the target row according to the contour line corresponding to the target row and the turning plate includes: Determining a marker with a deterministic included angle with the laying direction of the target row in the first sonar image according to the contour line corr