Guided Classification using CCV and Wavelet Features for Greenery and Non-Greenery Images

This paper focuses on finding the image features in both spatial and wavelet domain and applied that for greenery and non-greenery image classification. The spatial color feature called guided color coherence feature(GCCF) & guided distance feature in wavelet(GDFW) are calculated for the purpose of classification. The color coherence vector is constructed for all images and is used to find three GCCFs namely Delta_H, Delta_G & Delta_I . For finding guided distance feature in wavelet (GDFW), the input images are decomposed into 16 blocks using 2-dimensional discrete wavelet transform. Then wavelet image feature vector is formed by calculating mean and variance in each block. The difference between guide wavelet image and input wavelet image is named as guided distance feature in wavelet(GDFW) is calculated. Here one or more guide images are used for finding these guided features. These parameters are given in different combinations as input to the adaptive neuro fuzzy inference system(ANFIS). The performance of system is illustrated with a set of 600 images containing both greenery and non-greenery images.