Dear Dr. Chavan I suggest to read http://www.isprs.org/proceedings/XXXVII/congress/1_pdf/63.pdf. It is different from medium resolution 2-D remote sensing. Similar observations on high biodiversity was observed by Lee et al. (2007) who carried out Ordination analyses which reveals the significance of forested habitats with dense native vegetation cover. Lee, T. M., Sodhi, N. S., & Prawiradilaga, D. M. (2007). The importance of protected areas for the forest and endemic avifauna of Sulawesi (Indonesia). Ecological Applications, 17(6), 1727-1741.

there are so many ways to perform this based on experience.

but simple way is go for ndvi and mask forest areathen cut the fcc using this mask tool. run unsupervised classification giving 50 class or more.then based on swiping the classified out put on fcc fix the threshold of canopy class. 

Thank Vipin. I need to specify my question. I am actually looking for canopy variance for birds and scale smaller than that of LISS IV also needs to be considered. Additionally, the variance; vertical, horizontal and within canopy needs to be estimated.So, I am looking for field techniques to measure canopy variance.

Thanks for your response Vipin.