With the continuous development of artificial intelligence and computer vision technology, autonomous vehicles have developed rapidly. Although self-driving vehicles have achieved good results in normal environments, driving in adverse weather can still pose a challenge to driving safety. To improve the detection ability of self-driving vehicles in harsh environments, we first construct a new color levels offset compensation model to perform adaptive color levels correction on images, which can effectively improve the clarity of targets in adverse weather and facilitate the detection and recognition of targets. Then, we compare several common one-stage target detection algorithms and improve on the best-performing YOLOv5 algorithm. We optimize the parameters of the Backbone of the YOLOv5 algorithm by increasing the number of model parameters and incorporating the Transformer and CBAM into the YOLOv5 algorithm. At the same time, we use the loss function of EIOU to replace the loss function of the original CIOU. Finally, through the ablation experiment comparison, the improved algorithm improves the detection rate of the targets, with the mAP reaching 94.7% and the FPS being 199.86.

Convolutional neural networks have achieved good results in target detection in many application scenarios, but convolutional neural networks still face great challenges when facing scenarios with small target sizes and complex background environments. To solve the problem of low accuracy of infrared weak target detection in complex scenes, and considering the real-time requirements of the detection task, we choose the YOLOv5s target detection algorithm for improvement. We add the Bottleneck Transformer structure and CoordConv to the network to optimize the model parameters and improve the performance of the detection network. Meanwhile, a two-dimensional Gaussian distribution is used to describe the importance of pixel points in the target frame, and the normalized Guassian Wasserstein distance (NWD) is used to measure the similarity between the prediction frame and the true frame to characterize the loss function of weak targets, which will help highlight the targets with flat positional deviation transformation and improve the detection accuracy. Finally, through experimental verification, compared with other mainstream detection algorithms, the improved algorithm in this paper significantly improves the target detection accuracy, with the mAP reaching 96.7 percent, which is 2.2 percentage points higher compared with Yolov5s.

To improve the detection ability of infrared dim and small targets in complex cloud backgrounds, we present a novel method based on motion energy estimation. First, according to the characteristic of the gradient between the target and the background, we propose a new diffusion function and combine with the gradient difference to complete the background suppression. Then an improved high-order correlation algorithm is proposed to enhance the target signal through the motion correlation properties of the target, which build the gradient difference discrimination model between the target and four adjacent directions to establish the region of interest model, then realizes the high-order energy enhancement of the region of interest and further eliminate the clutters at the same time. Finally, the motion model of the target in the inter-frame spatial-temporal domain is constructed to estimate the motion direction of the target, then realizes the target energy accumulation along the motion direction and obtains the maximum energy of the target to detect the real target. In the experiment, we select four scenes for testing, the experimental results show that the proposed method has an excellent detection performance compared with six traditional detection algorithms in infrared dim and small target detection.

When faced with complex scenes containing strong edge contours and noise, there are still more background residuals in the detection results of traditional algorithms, leading to a high false alarm rate. To solve the above problems, we propose an infrared dim and small target detection method that combines local feature prior and tensor train nuclear norm(TTNN). To suppress the strong edge contours, we first establish a background edge contour suppression function based on the structure tensor. Secondly, we propose a multi-frame density peak search algorithm to obtain local feature information by combining the features associated with multiple contiguous frames of the target. Then we use the local feature information and reweighting strategy to constrain the sparse components of the target signal, and describe the background low-rank components by the tensor train nuclear norm. Finally, we separate the target image from the background image using the alternating direction multiplier method. As compared with eight advanced algorithms, the method in this paper has a better background strong edge contour and a stronger noise suppression.

Dim and small target detection plays an important role in infrared target recognition systems. In this paper, we present a dim and small target detection algorithm based on improved bilateral filtering and Gaussian motion probability estimation, aiming to improve the detection efficiency of the detection system. First, a bilateral filtering algorithm based on image patch analysis is proposed to complete the background modeling, compare with single pixel, image patch contains more neighborhood information. Then, we use the Gaussian process combining the target position of consecutive n frames to predict the target position of the (n + 1)th frame, and the target energy is accumulated along the trajectory direction at the same time. Finally, we construct the grayscale probability model to realize the multi-frame correlation detection, which combining the grayscale features and the motion characteristics of the target. Six scenes and eleven comparison algorithms are selected for experiments, experimental results show the effectiveness and robustness of the proposed algorithm.

Infrared target detection holds significant importance in automatic target search and tracking system under complex background. The conventional structural tensor models have not harnessed the full potential of spatio-temporal domain information in sequence scenes, and the strong edge contours in the image often lead to false alarms. In order to tackle this problem, we propose an improved twist tensor model based on the optimization of background constraints. Firstly, we propose a diffusion function according to the gradient difference between the target and backgrounds, which preserves the target signal to a great extent. Secondly, the spatio-temporal information of the sequence images is used to construct the twist tensor objective constraint optimization function, and the improved twist tensor effectively distinguishes the sparse components of the target and the low-rank components of the background. Finally, the optimized model is solved using ADMM to obtain the final target signal. Eight sequence images and nine comparison methods are performed for experimental validation, experiment results demonstrate that the proposed algorithm can effectively suppress the complex background while retaining the target signal well.