Sea surface remote sensing image ship detection method based on a feature pyramid

Abstract

The invention discloses a sea surface remote sensing image ship detection method based on a characteristic pyramid. The method comprises the steps of segmenting a remote sensing image of a sea surfacebackground ship to be detected into small images through a sliding window mechanism; screening out small images which may contain ships from the small images obtained by segmentation through a ship classification pre-inspection network; detecting the screened small graphs which may contain the ship through a target detection network based on a feature pyramid to obtain a small graph detection result; and fusing spliced small image detection results to obtain a final detection result of the sea surface background remote sensing image. According to the size of the target, the target is dividedto a specific coding position, so that the specific position processes the target with the corresponding size in the coding matrix, and the network convergence speed in the training process is increased. And an up-sampling structure of the feature pyramid uses a hole convolution operation, so that the receptive field is improved under the condition that the size of the feature map is not changed.The loss function only calculates the loss of the part corresponding to the target scale, simplifies the network learning task, and accelerates the network convergence in the training process.

Description

technical field [0001] The invention belongs to the technical field of image recognition, and more specifically relates to a method for detecting ships in sea surface remote sensing images based on feature pyramids. Background technique [0002] As an important tool for utilizing and developing the ocean, ships have drawn more and more attention to ship detection. In terms of civilian use, ship detection has important uses in fishery monitoring and management, port ship navigation management control, marine pollution monitoring, etc.; in military use, ship detection is used to strengthen sea area supervision, monitor illegal ships and monitor wartime sea surface to detect enemy ships etc. have broad application prospects. For large-scale remote sensing image data in the ocean background acquired by high-altitude reconnaissance platforms such as satellite remote sensing imaging, relying solely on manual interpretation has a large workload, high repeatability, and low efficie...

AI Technical Summary

Problems solved by technology

[0004] In view of the defects of the prior art, the purpose of the present invention is to solve the technical problems that the existing target detection network in the prior art cannot directly process large-format images, and the existing detection algorithm is slow, the network structure is complex, and the detection accuracy of smaller targets is low.

Images