This paper describes a novel method for correcting a color cast in an image which contains RGB and near-infrared crosstalk. We name this type of images as four-band images. The algorithm is based on the back propagation neural network. When a four-band image is put into the algorithm, it will use a trained weight matrix to transform this image into the normal one. This matrix is obtained by a BP algorithm which is used to find the implicit relationship between the input image and the corresponding target image. The size of the matrix is determined by the number of hidden layers of the neural network and the number of rows of the pixel matrix of the input and output images. In this paper, we acquire the weight matrix of the input layer to the hidden layer with a size of 506*486 and the weight matrix of the hidden layer to the output layer of 486*506. The camera used herein is a slightly modified CMOS sensor that replaces the IR-Cut filter with an 850 nm bimodal filter. The dataset used in this paper has five pairs of one-to-one correspondence images. One of them is a three-band image with normal color using the original IR-Cut filter, and the other is a four-band image taken by the modified CMOS sensor. The proposed method has been tuned and tested with positive results in this dataset.
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.