Image processing method

Abstract

The invention provides an image processing method. The method comprises the following steps that 1, for each pixel point of an image, grayscale frequency and coordinate information of the pixel pointare recorded; 2, the grayscales of all the pixel points in the image are clustered according to the gray level types to form a spatial gray level area position distribution diagram Ti with different gray level types as characteristics; 3, all gray level blocks in the gray level area position distribution diagram with different gray level types are screened, and target blocks are selected; the target blocks which are found from the gray level area position distribution diagram with different gray level types are combined, identified and subjected to image reconstruction. Compared with originaldeep leaning image identification, the image processing method has the advantages that another way is provided, a grayscale standard is adopted for conducting block division on the image, the restrictions of the learning capacity and the variable target edge gradient are removed, the data processing speed is greatly increased, the identification validity is greatly improved, the requirement for hardware is remarkably reduced, the high universality is achieved, and the difficult problems of complex environment, low contrast degree and tracking identification for a small target are solved.

Description

technical field [0001] The invention belongs to the technical field of software and relates to an image processing method. Background technique [0002] At present, the key to intelligent equipment is the cognition of the visual scene. Only when the cognition of the visual scene is accurate can one reasonably control one's own behavior. Currently, the most widely used technology for visual scene cognition is deep learning technology. [0003] Deep learning is a method based on representation learning of data in machine learning. Observations (such as an image) can be represented in a variety of ways, such as a vector of intensity values ​​for each pixel, or more abstractly as a series of edges, regions of a specific shape, etc. Instead, it is easier to learn tasks from examples (e.g., face recognition or facial expression recognition) using certain representations. The advantage of deep learning is to use unsupervised or semi-supervised feature learning and hierarchical fe...

AI Technical Summary

Problems solved by technology

In layman's terms, once a layer of training cannot be effectively used by the next layer, then this learning is invalid. Therefore, when deep learning processes visual images, due to the above shortcomings, the learning libraries required for deep learning and the support for these librari

Images