A complex remote sensing land environment small sample and small target rapid detection and identification method

Abstract

The invention discloses a complex remote sensing land environment small sample and small target rapid detection and identification method, and the method comprises the steps of constructing a detection and identification network for a complex remote sensing land environment vehicle target based on an improved Faster R-CNN convolutional neural network architecture; carrying out certain transformation and disturbance expansion on training data and carrying out repeated training on negative samples and difficult-to-distinguish samples, so that a network can fully learn the change of a target while the training data volume is increased, and the problems of weak model generalization ability and poor precision caused by small sample data volume are solved; adding small target features and mining difficult sample information, so that the problems that Faster R-CNN is poor in small target detection effect, high in false alarm rate and low in recognition precision are solved; the RPN and the Fast R-CNN sharing the same five-layer convolutional neural network, and adjusting and optimizing network model parameters, so that the whole detection process only needs to complete a series of convolution operations to complete the detection and identification process, and the operation time is reduced.

Description

technical field [0001] The invention belongs to the technical field of machine learning and deep learning, and specifically relates to a rapid detection and identification method for small samples and small targets in complex remote sensing land environments. Background technique [0002] The detection and extraction research institutions of land targets in satellite images are concentrated in Europe and the Asia-Pacific region. Most of them use radar data or high-resolution data to carry out related scientific research activities. Their basic research is solid, and the examples are also very rich. The research conducted is quite comprehensive. Whether it is in the model building and algorithm development of land target extraction, or in the extraction and classification of land targets in different spatial distribution forms and different motion states, there are rich research results. [0003] Foreign researchers started early on the application research of traffic data co...

AI Technical Summary

Problems solved by technology

[0005] At present, domestic research on the extraction of land targets in satellite remote sensing images mostly focuses on the extraction of large-scale land such as urban land and buildings. There are not too many studies on the detection and speed extraction of tiny land targets.

[0006] In addition, the biggest problem existing in the existing land target extraction methods of remote sensing images is how to extract dark and light vehicles in complex backgrounds.

The simple background land target extraction based on threshold and geometric shape has already had certain research results, and some researchers have also begun to try the object-oriented land target extraction method. In terms of machine learning, the neural network learning method has already had some research results, but There is still a lot of room for improvement and breakthrough, which needs further research

Images