• P.DOLPHIN DEVI

Multisensor fusion is of abundantsignificance in Earth observation related applications. For instance, multispectral images (MSIs) deliverfull spectral data while light detection and ranging (LiDAR) data deliverheight information, and using MSI and LiDAR data together can attainimproved classification performance. In this work, an unsupervised feature extraction model, calledas N-gram along with classifier called as patch-to-patch convolutional neural network (N-Gram: PToP CNN), is proposed for collaborative classification of mutispectral and LiDAR data. Sparse coding, or sparse dictionary learning, is an unsupervised learning algorithm, and is skilled of mining features based simply on how well those features can be familiar torebuild the original image. With respect to imagepatches, we acquire sparse dictionaries for n-grams, constant sequences of bytes, of various sizes.More specific, a three-tower PToP mapping is first established to find an accurate illustration from MSI to LiDAR data, aiming at merging N-gram features between two different images. Then, by combining hidden layers of the designed PToP CNN, extracted features are projected to retain deeply fused characteristics. Hence, features from various hidden layers are merged into a stacked vector and provide into three fully inter connected layers. To prove the efficiency of the proposed classification structure, experiments are implemented on benchmark remote sensing data sets. The experimental results prove that the proposed method offersgreater performance when compared with some state-of-the-art classifiers, such as two-branch CNN and context CNN.

Mutispectral image, classification,. N-gram, CNN, patch to patch

"N-Gram: PToPCNN : An Improved Image Classification Using Patch-To-Patch Convolutional Neural Network With Sparse Based N-Gram Feature Extraction Method", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 4, page no.461-465, April-2019, Available :http://www.jetir.org/papers/JETIR1904E72.pdf

"N-Gram: PToPCNN : An Improved Image Classification Using Patch-To-Patch Convolutional Neural Network With Sparse Based N-Gram Feature Extraction Method", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 4, page no. pp461-465, April-2019, Available at : http://www.jetir.org/papers/JETIR1904E72.pdf