Target detecting method, target detecting equipment and image pickup equipment

Abstract

The invention relates to a target detecting method, target detecting equipment and image pickup equipment. The target detecting method comprises the following steps of storing a universal model, i.e. storing the universal model for a special target type off line in advance, wherein the universal model describes multiple components stored in each single target of the special target type; receiving sample images, i.e. receiving inputs of one or more sample images, wherein each sample image comprises the same inquiry target, and the inquiry target is the target of the special target type; building a target detector, i.e. using the universal model and the one or more sample images to build the detector special for the inquiry target; detecting targets, i.e. using the built detector special for the inquiry target to detect the inquiry target from the target images. According to the target detecting method, the target detecting equipment and the image pickup equipment, various targets of the special target type can be highly flexibly and precisely detected.

Description

technical field [0001] The present invention relates to an object detection method, an object detection device, and an image pickup device including the object detection device. Background technique [0002] The analysis of visual objects in an image is important in processes such as object recognition, image retrieval, and image registration. Such processing is involved in many fields of application, such as surveillance (for security purposes), video forensics, and medical image analysis for computer-aided diagnosis, among others. In particular, object recognition has attracted increasing attention due to the growing demand for developing systems relevant to the real world. [0003] In conventional object recognition methods, a learning / training based classifier is used, for which a learning / training process of classifier parameters is indispensable. However, a classifier obtained through a learning / training process using specific samples of one object (eg, dog) may not ...

AI Technical Summary

Problems solved by technology

[0006] However, the inventors of the present invention have found that the methods disclosed in the above patent documents are not good enough because the specific objects that can be detected are still limited, and only the specific objects registered by the user in advance can be detected

Images