We consider the estimation of animal density by distance sampling line transect survey when the data of sighting distances are contaminated with measurement errors. The presence of measurement errors may result in a substantial bias on the estimation of animal density. This study investigates the effects of measurement errors while applying the nonparametric kernel density estimation. With regards to the typical additive and a general multiplicative measurement error models, we accordingly develop corrections to adjust the biases caused by measurement errors. A simulation study was carried out for comparing with other parametric approaches. An extension to the distance sampling point transect surveys is addressed as well.