Density-based image processing method and device and equipment

Abstract

The invention provides a density-based image processing method and device and equipment with an image processing function. The density-based image processing method comprises the following steps of: carrying out edge extraction processing on a to-be-processed image; convoluting the image which undergoes the edge extraction processing via a convolution step length by using a target convolution kernel, so as to obtain an edge density point space of the image which undergoes the edge extraction processing; screening points in the edge density point space by using a density threshold value; obtaining a target connected domain in the screened edge density point space; and determining a boundary of a target object in the image according to the target connected domain. Through above technical scheme, non-target objects in images can be correctly removed, and the image identification precision and speed can be improved, so that the image identification correctness and instantaneity are ensured.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular, to a density-based image processing method, a density-based image processing device and a device with image processing functions. Background technique [0002] In recent years, with the rapid development of artificial intelligence and big data technology, more and more products have begun to develop towards intelligence. Compared with non-intelligent products, intelligent products have more powerful functions and more comfortable user experience. Among the many directions of intelligence, image recognition is an important field of intelligence. A complete image recognition system uses images as input information, recognizes objects in the image through different methods, and finally outputs the recognition results. However, in the recognition image, due to the existence of many complex backgrounds and non-target objects, it will have a great impact on the recogniti...

AI Technical Summary

Problems solved by technology

However, in the recognition image, due to the existence of many complex backgrounds and non-target objects, it will have a great impact on the recognition results. Therefore, before recognizing the image, it is necessary to segment the image to extract a pure image and remove complex backgrounds and objects. Non-target

[0003] The current mainstream image segmentation methods mostly use threshold segmentation, color gradient segmentation, and edge segmentation. According to the above principles, a large number of universal segmentation methods are derived, but the above algorithms do not perform well in specific subdivision scenes. , and the segmentation takes a long time

Especially for images with complex and changeable textures and large color changes, the segmentation effect in the prior art is not good

[0004] The current image segmentation methods are mainly based on two methods, which are edge-based and region-based segmentation methods. Among them, if there are complex and irregular cross-textures in the image, it is impossible to accurately remove non-target objects in the image, while region-based Due to the diversity of image content, the advanced segmentation algorithm often leads to segmentation errors, and the segmentation time is too long and the real-time performance is poor, so non-target objects cannot be automatically eliminated.

Images