Image registration is a typical problem and technical difficulty in the research field of image processing, which can be used in infrared image fusion, image Mosaic, image segmentation, super-resolution reconstruction and other directions. On the basis of infrared image registration, considering the difference of infrared image, the effective accuracy of registration is set. The algorithm in this paper was initially used in super-resolution reconstruction, but it was redesigned in the registration process due to the need to reduce the computing cost in hardware implementation. The algorithm is improved on the basis of the original flownetS, and the cost of the calculate is reduced by 89% when the number of convolutions layers and the receptive field of the improved network are the same. There is no need for pooling layer, because pooling provides greater receptive field while reducing resolution, which will lead to spatial information loss and data loss. Dilate convolution convolution can avoid down-sampling and provide a larger sense field under the premise of the same amount of computation. Set different Dilation rates, receptive fields will be different, and there will be multi-scale information. This method based on deep learning uses the processing method of dilated convolution to greatly reduce the cost of network computing and provide a convenient channel for the hardware implementation of the algorithm. Moreover, the algorithm also achieves excellent results on the development board RK3399Pro. Combined with infrared image, this paper also demonstrates the difference between infrared image in the registration of effective accuracy and visible light, and also analyzes the target at different speed, the minimum network structure required. Finally, an appropriate ratio was selected through experimental attempts to ensure stable accuracy in the conventional moving target.
Infrared imaging system compared with visible light imaging system, infrared imaging system can operate in all-weather conditions with high anti-interference ability and the ability to penetrate smoke and haze. The current infrared technology still has the disadvantage of low signal-to-noise ratio and lack high frequency detail information. The most direct way to improve the infrared imaging system is to improve the hardware design by increasing the size of the photoreceptor and the size of the image source, but the manufacturing process is complicated, the cost is high, and the improvement effect is limited. With the development of the field of computer vision, deep learning has become the most effective solution for super-resolution reconstruction[6]. Therefore, in this paper, a Video frame Infrared image super-resolution reconstruction method IVSR (Infrared Video super-resolution) is designed based on the deep learning method and the dual-path operation mode combining the single-frame and multi-frame super-resolution reconstruction algorithms. The innovation of IVSR net lies in that after the optical flow module extracts the inter-frame information of each frame and the current frame, the sub-pixel convolution layer of the multi-frame image fusion module can effectively utilize the sub-pixel information. IVSR can be regarded as two reconstruction processes, the first stage is the output of high-resolution information from the optical flow motion estimation module to the projection module, and the second stage integrates the high-resolution information of each projection module for fusion reconstruction. This method can effectively improve the visual effect of IVSR reconstruction, effectively solve the problem of poor reconstruction quality due to the lack of high frequency detail information. Compared with traditional reconstruction algorithms and other typical deep learning algorithms, reconstructed images of IVSR are more exquisite, with more prominent high-frequency details, no image distortion and significant advantages in objective evaluation indicators.
Infrared images typically contain obvious dark-corner noise. It is a challenging task to eliminate such noise with the acceptable computation overhead and time overhead. In this paper, we introduce an effective dark-corner noise removal algorithm consists of two consecutive processing procedures. Firstly, in order to effectively filter dark-corner noise with as few as frames of infrared images, the proposed algorithm accumulates the low-frequency pixels during the several different frames of infrared images and eliminates the dark-corner noise by subtracting this parameter from the original infrared image. Then, this algorithm sets several detection windows for dark-corner noise to obtain another additive correction parameter and subtract this parameter from the original infrared image. We demonstrate the effectiveness of our algorithm from experimental perspective.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.