Object detection method based on depth image and implementing device thereof

Abstract

The invention discloses an object detection method based on a depth image and an implementing device of the object detection method. The object detection method comprises the steps of a training process and a detection process, an image is read from a camera and serves as a data source, an object is segmented on the image according to the distance between the object to be detected and the camera, characteristics of the appearance and the shape of the object are calculated according to the segmentation, and a classifier aiming at the object is trained according to the calculated characteristics of the appearance and the shape of the object; the classifier obtained in the training process is utilized to judge the characteristics of the appearance and the shape inside a scanning window, whether a target object exists inside the scanning window or not is determined, detection results of layers of an mage pyramid are infused, and a final detection result is obtained. The depth image is not affected by changes of illumination or complex backgrounds easily, and accordingly the false detection number is reduced; the speed of detection is improved remarkably, and the real-time performance of the system is ensured.

Description

Technical field [0001] The invention involves mode recognition and human -computer interaction fields. More specifically, it involves an object detection method based on deep images and its implementation device. Background technique [0002] Image -based object detection refers to the process of determining whether there is a certain or several types of objects in a given image, and give the object's specific position in the image in the image.In the field of mode recognition and human -computer interaction, commonly tested objects include: vehicles, human body, face, hand, etc.By determining the size, location, and motion of these objects, computers can respond to specific visual goals or people's behavior, and finally realize the purpose of human -computer interaction or computer automatic processing recognition. [0003] Two key steps during object testing: feature selection and classifier learning.The problem that the feature selection needs to be solved is: for the current ...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem that the number of detection windows is too large in the above object detection technology, which limits the application of complex features with a large amount of calculation, and proposes a Object detection method and implementation device based on depth image

Images