Higher Order Dynamic Conditional Random Fields Ensemble for Crop Type Classification in Radar Images

ABSTRACT:

The joint sparse representation (JSR)-based classifier assumes that pixels in a local window can be jointly and sparsely represented by a dictionary constructed by the training samples. The class label of each pixel can be decided according to the representation residual. However, once the local window of each pixel includes pixels from different classes, the performance of the JSR classifier may be seriously decreased. Since correlation coefficient (CC) is able to measure the spectral similarity among different pixels efficiently, this letter proposes a new classification method via fusing CC and JSR, which attempts to use the within-class similarity between training and test samples while decreasing the between-class interference. First, the CCs among the training and test samples are calculated. Then, the JSR-based classifier is used to obtain the representation residuals of different pixels. Finally, a regularization parameter k is introduced to achieve the balance between the JSR and the CC. Experimental results obtained on the Indian Pines data set demonstrate the competitive performance of the proposed approach with respect to other widely used classifiers.

PROJECT OUTPUT VIDEO:

EXISTING SYSTEM:

• In existing paper, they presented a method for the multitemporal and contextual classification of georeferenced optical remote sensing images acquired at different epochs and having different geometrical resolutions.

• The method is based on Conditional Random Fields (CRFs) for contextual classification. The CRF model is expanded by temporal interaction terms that link neighboring epochs via transition probabilities between different classes.

• In order to be able to deal with data of different resolution, the class structure at different epochs may vary with the resolution. The goal of the multitemporal classification is an improved classification performance at all individual epochs, but also the detection of land-cover changes, possibly using lower resolution data. 

DISADVANTAGES OF EXISTING SYSTEM:

• In existing, noted that incorrect determination of transition matrix could lead to erroneous transfer of information to other epochs consequently reducing classification accuracy.

PROPOSED SYSTEM:

In this paper, it is found that the correlation coefficient (CC) can recognize those pixels with strong relationships effectively. Therefore, this letter proposes a classification method by combining CC and JSR (CCJSR) for classification. It can be accomplished by three main steps. In the first step, a test sample can be linear represented by the atoms in an over complete dictionary and sparse vectors. In this step, JSR is used to produce the residual for every class. In the second step, CC is used to calculate the degree of similarity between the training and test samples.

In the last step, a decision function is used for classification based on the residual of JSR and the degree of correlation. The proposed method combines two major factors, i.e., spectral similarity and local spatial consistency, for crop type classification.

ADVANTAGES OF PROPOSED SYSTEM:

• Our max F1-score model performed better than other existing ensembles and also compared with stacking multitemporal images for classification.

• In addition, the designed technique detected changes in crop parcels. This presents the possibility of using it to monitor changes in agricultural areas, i.e., due to farm management or natural disasters like wind destruction of crops.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

• System           : Pentium Dual Core.

• Hard Disk      : 120 GB.

• Monitor          : 15’’ LED

• Input Devices : Keyboard, Mouse

• Ram                : 1GB.

SOFTWARE REQUIREMENTS:

• Operating system : Windows 7.

• Coding Language : MATLAB

• Tool                      : MATLAB R2013A

REFERENCE:

Benson Kipkemboi Kenduiywo, Damian Bargiel, and Uwe Soergel, Member, IEEE, “Higher Order Dynamic Conditional Random Fields Ensemble for Crop Type Classification in Radar Images”, IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2017.