Our activities

We select, experiment with and implement machine learning technologies. Only advanced algorithms are able to handle the enormous amounts of data which is acquired and derived every single day.

Artificial neural networks applied in the classification of point clouds

OBJECTIVE:

Reducing the amount of manual work needed during classification of the airborne laser scanning point clouds. The currently used automatic classification algorithms of point clouds do not produce satisfactory results, particularly in respect of urban area analyses. The results of automatic classification require labour-intensive manual correction during which the classification errors are corrected and the classification of temporary structures/buildings and urban infrastructure structures/buildings is carried out from scratch. Improvement in the automatic classification efficiency will allow for reducing the time dedicated to manual correction of the classification, as well as for improving the quality of the resulting products.

RESULT:

The developed method of processing point clouds into matrixes allowed for the use of a convolutional neural network within the context of the point cloud classification. A high level of the classification efficiency was achieved by using an extensive set of correctly classified data for training the artificial neural network. The classified point cloud, obtained in this manner, exceeds in terms of quality the classification obtained by means of the previous commonly applied methods in many aspects. Not only did it allow for reducing the labour intensity dedicated to manual correction of the classification, but it also resulted in improving the resulting quality of the classification.

CUSTOMER:

Contractor responsible for preparing analyses / products based on LiDAR data

OPEGIEKA Sp. z o.o.

Click here for directions.

Head Office

Aleja Tysiąclecia 11

82-300 Elbląg

phone: +48 (55) 237 60 00

fax: +48 (55) 237 60 01

E-mail address:

lab@opegieka.pl

Branch in Warsaw

ul. Grzybowska 80/82 lok. 700

00-844 Warszawa