Object detecting with 1D range sensors
View Patent ↗Moving objects are classified based on maximum margin classification and discriminative probabilistic sequential modeling of range data acquired by a scanner with a set of one or more 1D laser line scanner. The range data in the form of 2D images is pre-processed and then classified. The classifier is composed of appearance classifiers, sequence classifiers with different inference techniques, and state machine enforcement of a structure of the objects.
1. A method for classifying an object in a scene, comprising the steps of:
preprocessing a sequence of images, wherein each image is acquired of the object in the scene by a scanner, wherein the scanner includes a 1D laser line sensor, and each image includes columns of pixels, and each pixel has an associated depth value such that each image is a range image, wherein the preprocessing further comprises;
denoising each image in the sequence;
removing background pixels from each image;
projecting, in 3D, each image to a 3D world coordinate system;
correcting depth values;
extracting features; and
classifying the sequence of images, wherein the classifying further comprises:
applying an appearance classifier to the features to obtain labels;
applying a sequence classifier to smooth the labels; and
enforcing a structure of the object to determine a class of the object, wherein the steps are performed in a processor.
2. The method of claim 1 , wherein 1D laser line scanner scans a side of the object.
3. The method of claim 1 , wherein the denoising uses a 2D median filter.
4. The method of claim 1 , wherein a background of the scene is modeled with a Gaussian distribution for each pixel.
5. The method of claim 1 , wherein a ground plane in the scene is estimated using a RANdom SAmple Consensus (RANSAC) process.
6. The method of claim 1 , wherein noisy range measurements on surfaces of the object are corrected by fitting vertical planes to each column.
7. The method of claim 1 , wherein the features are scale invariant.
8. The method of claim 1 , wherein the features are binary height maps equal to a quantized side view of the 3D projection.
9. The method of claim 1 , wherein the scanner is mounted on a pole near the object.
10. The method of claim 1 , wherein the labels are determined by evaluating with the appearance classifier by a sliding window technique along a time axis.
11. The method of claim 1 , wherein the appearance classifier is a support vector machine.
12. The method of claim 1 , wherein the outputs of the appearance classifier are smoothed with the sequence classifier.
13. The method of claim 1 , wherein the sequence classifier uses a conditional random field model or a hidden Markov model.
14. The method of claim 1 , wherein the structure enforcing converts the labels to labels of a most similar valid object model defined in a object grammar.
15. The method of claim 1 , wherein the structure enforcing uses an error correcting regular grammar parser.