Feature Analysis In Image Processing

1. INTRODUCTION

Feature extraction in image processing is a method of transforming large redundant data into a reduced data representation. Transforming the input data into the set of features is called feature extraction.

Image analysis involves investigation of the image data for a specific application. Normally, the raw data of a set of images is analyzed to gain insight into what is happening with the images and how they can be used to extract desired information. In image processing and pattern recognition, feature extraction is an important step, which is a special form of dimensionality reduction. When the input data is too large to be processed and suspected to be redundant then the data is transformed into a reduced set of feature

Being motivated by this, proposed CLBP, an extension of the original LBP operator that assigns a 2P-bit code to the center pixel based on the gray values of a local neighborhood comprising P neighbors. Unlike the LBP operator that employs one bit for each neighbor to represent only the sign of the difference between the center and the corresponding neighbor gray values, the proposed method uses two bits for each neighbor in order to encode the sign as well as the magnitude information of the difference between the center and the neighbor gray values. Here, the first bit represents the sign of the difference between the center and the corresponding neighbor gray values like the basic LBP encoding. The other bit is used to encode the magnitude of the difference with respect to a threshold value, which is the average magnitude Mavg of the difference between the center and the neighbor gray values in the local neighborhood of interest. The CLBP operator sets this bit to 1 if the magnitude of the difference between the center and the corresponding neighbor is greater than the threshold Mavg. Otherwise, it is set to

As 16-bit codes are utilized to character the pixels, the number of available binary patterns is 216. To reduce the number of features, This expected to consider less number of neighbors while forming the binary patterns. Thus, the feature vector of length can be reduced by discarding some degree of neighborhood information. In this paper, we have presented a various approach where all the CLBP binary patterns are further separation into two sub-CLBP patterns. Each sub-CLBP pattern is obtained by concatenating the bit conducts complementary to P/2 neighbors, where P is the number of neighbors. The 16-bit CLBP pattern is separation into two 8-bit sub-CLBP patterns, where the first one sub-CLBP1 is obtained by concatenating the bit values complementary to the neighbors in the various direction (e.g. north, east, south, and west directions), respectively and the second sub-CLBP pattern sub-CLBP2 is obtained by concatenating the bit values complementary to the neighbors in the some directions (e.g. north-east, south-east, south-west, and north-west directions), respectively. The two sub-CLBP patterns are treated as separate binary codes and joined during the feature vector