Target detection and identification method under rain and snow weather conditions

Abstract

The invention discloses a target detection and identification method under rain and snow weather conditions, which comprises the following steps: firstly, acquiring to-be-detected rain and snow weather background image data containing a target, and constructing a training data set under the corresponding rain and snow weather conditions through a fuzzy filter; optimizing a separation model of a rain and snow layer and a background layer, and respectively processing a rain area and a rain-free area so as to weaken detail loss of the rain and snow-free area; building a context expansion rain removal network based on the scene information so as to restore background images under different rain and snow degrees; building a convolutional neural network based on local feature learning, and improving the target recognition rate; taking the rain and snow removed image as the input of an optimized target detection network model, extracting an interested area in the image data by the target detection network model, and outputting a target category; the rain and snow removal target detection model provided by the invention has good generalization and universality, and can be widely applied to high-precision automatic identification and detection of various targets under different rain and snow degrees and other actual scenes.

Description

Technical field [0001] The present invention belongs to the field of image processing, and more particularly to a target detection recognition method. Background technique [0002] The complicated weather conditions such as rain and snow will greatly affect the performance of the monitoring sensor, resulting in an increase in image noise, detail loss, contrast decline, and target identification detection as a pretreatment part of most computer visual applications. The identification effect is inevitably subject to different levels of rain and snow conditions. Therefore, the rain and snow interference in the input image can be quickly and accurately removed, so that the accuracy of improving the target identification detection ground is extremely value in various complex weather conditions. The traditional to rain and snow method is to get a prior information from the background and rain lines, mainly describe the background layer and the rain snow layer by constructing the mist, ...

AI Technical Summary

Problems solved by technology

Due to the lack of pertinence of the area selection algorithm, the selected area is not necessarily conducive to feature detection, so the efficiency is not high, and the extracted target features are likely to be ineffective due to the complexity of the target.

To sum up, the traditional detection method of rain and snow removal target recognition is easily affected by the degree of rain and snow and the algorithm improvement space is limited. It is necessary to combine the neural network algorithm to improve the accuracy of target recognition in rain and snow weather.

Images