The application discloses a kind of based on image inverse
perspective transformation and realizes
water level detection method, belong to the technical field of
computer vision application, the method comprises the following steps: S1:
water level gauge detection obtains
water level gauge overall target area and water level gauge scale target area;Water surface segmentation result is also obtained to water surface segmentation simultaneously;S2: according to water level gauge scale detection frame and water level gauge detection frame, inclination is judged, and pixel distance is calculated;S3: pixel distance between water level gauge detection frame top and water
level line is used, pixel is combined with the real distance represented to calculate water level gauge real height.The present application makes full use of the application in the field of image of
computer vision, carries out water level gauge detection and water surface segmentation using
deep learning model, realizes water level gauge inclination correction in image by inverse
perspective transformation, obtains water
level line position by water level gauge and water surface segmentation result intersection, to obtain accurate water level value.