Remote sensing infrared satellite images have the characteristics of weak targets, insufficient contrast, and easy to be affected by the surrounding environment, such as clouds and fog, so it is a great challenge to detect weak and small targets in remote sensing. In this paper, we propose a detection method based on weak and small target enhancement, which uses a bidirectional histogram to improve the image contrast, and uses the infrared image dehazing algorithm with fog line dark primary color prior to preserve the pixel distribution of the infrared image to the greatest extent while enhancing its contrast and detail. In terms of the model, we introduce a simple and efficient weighted bidirectional feature pyramid network to optimize feature fusion, reduce redundant calculations while maintaining the detection ability of the model, and greatly reduce the memory occupation. The results show that the proposed method has achieved more competitive results than the current mainstream methods in dealing with the problem of infrared weak and small target detection, and in addition, due to the application of the weighted bidirectional feature pyramid network, the video memory is reduced by 43% while maintaining the competitive accuracy, which is of great practical significance.