Object classification method and object classification device

Abstract

The invention discloses an object classification method and an object classification device. The method includes the steps of obtaining characteristic vectors of each block in an input image of an object; establishing sparse vectors corresponding to the characteristic vector of each block based on a predefined dictionary, wherein the distribution of the nonzero elements of the sparse vectors indicates whether the corresponding blocks belong to an area where the object is located; combining the sparse vectors corresponding to the blocks that belong to the area where the object is located to obtain a merging vector indicating the area where the object is located; and determining the type of the object according to the merging vector.

Description

technical field [0001] The present invention relates generally to the field of information processing. Specifically, the present invention relates to a method and device capable of detecting, locating, and classifying objects contained in an image. Background technique [0002] In recent years, image-based object detection and analysis techniques have developed rapidly in various fields. For example, in the field of medical image processing, the pathological conditions of tissues and organs can be analyzed based on input images, for example, lung tumors can be detected and classified. [0003] The more popular processing method is based on convolutional neural network (CNN). CNN can automatically select appropriate features for extraction, and judge whether there is a target object in the image area and the specific location of the target object based on the extracted features. [0004] However, CNNs have some usage limitations. For example, CNN needs to process for a gi...

AI Technical Summary

Problems solved by technology

The size and shape of the tumor are uncertain. If a larger rectangular area is set, it may completely include the tumor, but it also introduces a considerable amount of non-tumor area, which will bring interference information to the classification of the tumor. , the noise

In addition, the increase in the amount of data will bring about a decrease in efficiency

If a smaller rectangular area is set, although the noise is reduced, the tumor may not be completely contained in the rectangular area, resulting in the loss of important information and affecting the tumor classification results, such as false detection and missed detection

Even in an ideal situation, the rectangular area happens to contain the complete tumor, but the tumor area is not rectangular, so non-tumor information will still be introduced, causing interference

Images