CN111582290B - Computer image recognition method - Google Patents

Abstract

The invention discloses a computer image identification method, which comprises the following steps: extracting color values of all pixel points in an image to be recognized, generating a color matrix, and establishing coordinates of all the pixel points according to the color matrix; respectively comparing the color value of each pixel point with the color values of the adjacent pixel points around the pixel point, extracting the coordinate of the pixel point in the color matrix when the difference value between the color value of the pixel point and the color value of at least one adjacent pixel point around the pixel point is not in a set color difference range, and repeating the process until all the pixel points are traversed; obtaining one or more functions according to all the extracted coordinates; and searching in the image library according to the obtained function, searching to obtain the name of the article and outputting. According to the method, the pixel points of the image to be recognized are extracted, the outline of the article is separated from the pixel points, the article is recognized according to the outline of the article, and finally the article in the image is output.

Description

A computer image recognition method

technical field

The invention relates to the field of computers, in particular to a computer image recognition method.

Background technique

Image recognition is a basic technical means of artificial intelligence. At present, image recognition is by extracting the features of each pixel in the image to be recognized, and comparing it with the features of each pixel in the existing image. When the similarity of point features is within the set range, the item in the existing image is output as the item in the image to be recognized. Such a recognition method needs to extract the pixels of each compared image and perform a large amount of comparison work, so that when the computer performs calculations, it will greatly increase the amount of calculated data, so that the computer can recognize the image. Very slowly.

Contents of the invention

The purpose of the present invention is to overcome the problems existing in the above-mentioned prior art, and to provide a computer image recognition method, by extracting the pixels of the image to be recognized, separating the outline of the item from the pixel, and then according to the outline of the item. The item is identified, and finally the item in the image is output.

For this reason, the invention provides a kind of computer image recognition method:

The color value of each pixel in the image to be recognized is extracted to generate a color matrix, and the coordinates of each pixel are established according to the color matrix.

Compare the color value of each pixel point with the color value of its surrounding adjacent pixels, when the difference between the color value of the pixel point and the color value of at least one surrounding pixel point is not within the set color difference range , extract the coordinates of the pixel in the color matrix, and repeat this process until all the pixels have been traversed.

One or more functions are obtained according to all the coordinates extracted above.

Search in the image library according to the function obtained above, find the name of the item and output it.

The image library is used to store the names of items and corresponding functions.

Further, when comparing the color value of a pixel with the color values of its surrounding adjacent pixels, when the difference between the color value of this pixel and the color values of all surrounding adjacent pixels is set When the color difference range of the pixel is obtained, the difference between the color value of the pixel and the color values of all adjacent pixels around it is obtained, and the average of all the differences is calculated, and the average is added to the pixel. The color value of the point and replace the original color value of the pixel point.

Further, when the difference between the extracted color value of the pixel point and the color value of at least one surrounding pixel point is not within the set color difference range, change the color value of the pixel point to the set color value .

Furthermore, when one or more functions are obtained through coordinates:

Choose a function from the image library.

All the extracted coordinates are verified in the function in turn, and when at least two coordinates satisfy the function, the function is output.

Repeat the above process until all the functions in the image library are traversed.

Furthermore, when one of the values in the coordinates is substituted into the corresponding position of the function, and the difference between the obtained value and the other value of the coordinates is within a set error range, the coordinates are deemed to satisfy the function.

Further, the image library includes a web crawler module, and the web crawler module is used to update the content stored in the image library.

A computer image recognition method provided by the invention has the following beneficial effects:

1. By extracting the pixels of the image to be recognized, the outline of the item is separated from the pixel, and then the item is identified according to the outline of the item, and finally the item in the image is output;

2. Split the outline of the item, and express the outline of each segment through a function, and determine the type of item according to the position characteristics of each segment of the function;

3. Take the average value of the surrounding pixels for the pixels of the pixel point setting range difference, so that when identifying the items in the image, the image will be smoothed at the same time, making the display effect of the image softer.

Description of drawings

Fig. 1 is a schematic block diagram of the overall flow of a computer image recognition method provided by the present invention;

Fig. 2 is a schematic block diagram of the process of obtaining one or more functions through coordinates in a computer image recognition method provided by the present invention.

Detailed ways

A number of specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiments.

In this application document, unspecified component models and structures are all prior art known to those skilled in the art, and those skilled in the art can set them according to the needs of the actual situation. In the embodiments of this application document No specific restrictions are made.

Example 1

This embodiment provides a computer image recognition method, and implements the present invention through the basic necessary technical features to solve the problems raised in the technical background section of this application document.

Specifically, as shown in Figure 1, an embodiment of the present invention provides a computer image recognition method, specifically as follows:

The color value of each pixel in the image to be recognized is extracted to generate a color matrix, and the coordinates of each pixel are established according to the color matrix. In the present invention, each pixel point in the image has its uniquely corresponding coordinates, and each pixel point also has its uniquely corresponding color value, so in the coordinate matrix corresponding to each pixel point of the image, each The coordinates of the pixel point will correspond to a color value one by one.

Compare the color value of each pixel point with the color value of its surrounding adjacent pixels, when the difference between the color value of the pixel point and the color value of at least one surrounding pixel point is not within the set color difference range , extract the coordinates of the pixel in the color matrix, and repeat this process until all the pixels have been traversed. When the difference between the color value of one of the pixel points and the color value of at least one adjacent pixel point is not within the set color difference range, it means that the position of the pixel point is the dividing line of the item in the image, so by By traversing each pixel in the image to be identified, all the pixels on the boundary line can be collected. Generally, the color difference range used is 0 to 40, that is, the color value of one pixel is adjacent to its surroundings by at least one If the difference between the color values of the pixel points is above 40, the pixel point will be marked as the pixel point on the boundary line.

One or more functions are obtained according to all the coordinates extracted above. A function composed of all the coordinates is obtained according to all the extracted coordinates, and the obtained function may include one or more functions.

Search in the image library according to the function obtained above, find the name of the item and output it. By searching in the database, the name of the corresponding item is output.

The image library is used to store the names of items and corresponding functions. The image library stores the names of the stored items and the corresponding functions one by one in the form of a list.

Example 2

This example is based on Example 1 and optimizes the implementation in Example 1, so that this example is more stable and has better performance during operation, but it is not limited to the one described in this example implementation.

Specifically, as shown in Figure 1-2, in this embodiment, when comparing the color value of a pixel with the color values of its surrounding adjacent pixels, when the color value of this pixel and all its surrounding adjacent When the difference between the color values of the pixel is within the set color difference range, the difference between the color value of the pixel and the color values of all adjacent pixels around it is respectively obtained, and all the color values are calculated. The average of the differences, and the average is added to the color value of the pixel and replaces the original color value of the pixel.

In this embodiment, when the item is identified, the color of the image can be smoothed at the same time, and the average value can be added to the color value of the pixel and replace the original color value of the pixel, so that Make the color value of each pixel maintain a smooth transition, so that when the user watches the image, the smoothness of the color can make the visual effect better.

In this embodiment, when the difference between the color value of the extracted pixel point and the color value of at least one surrounding pixel point is not within the set color difference range, the color value of the pixel point is changed to the set value specified color value. In this way, the boundary of the item can be marked with the set color value, so that when the user sees the image again, the item in it can be clearly seen, and it is easy to effectively distinguish when viewing.

In this embodiment, when one or more functions are obtained through coordinates, the following steps are specifically performed, but are not limited to one of the following steps:

First, a function is selected from the image library.

Then, all the extracted coordinates are verified in the function in turn, and when at least two coordinates satisfy the function, the function is output.

Repeat the above process until all the functions in the image library are traversed.

The above method is to use the extracted coordinates to verify all functions in the image library. When verifying, the two coordinate points of the same function must be adjacent to the surrounding coordinate points to verify the function. At the same time, at least two When the coordinates satisfy the function, it is determined that the verification is successful, and the function is output. By traversing every Haishu, one or more functions satisfied by all the extracted coordinate points can be obtained.

At the same time, in this embodiment, when one of the values in the coordinates is substituted into the corresponding position of the function, and the difference between the obtained value and the other value of the coordinates is within the set error range, the Coordinates satisfy this function. In this way, when verifying, the error is guaranteed to make the verification of the function more accurate, so that the recognition rate of the obtained items will be higher.

In this embodiment, the image library includes a web crawler module, and the web crawler module is used to update the content stored in the image library. In this way, the data stored in the image database can always be at the latest level, and the stored data is the latest data of the entire network, ensuring that when calling the database, the accuracy of object recognition is kept at the latest level.

The above disclosures are only a few specific embodiments of the present invention, however, the embodiments of the present invention are not limited thereto, and any changes conceivable by those skilled in the art shall fall within the protection scope of the present invention.

Claims (2)

1. A computer image recognition method is characterized in that:

extracting color values of all pixel points in an image to be recognized, generating a color matrix, and establishing coordinates of all pixel points according to the color matrix;

respectively comparing the color value of each pixel point with the color values of adjacent pixel points around the pixel point, extracting the coordinates of the pixel point in the color matrix when the difference value between the color value of the pixel point and the color value of at least one adjacent pixel point around the pixel point is not in a set color difference range, and repeating the process until all the pixel points are traversed;

obtaining one or more functions according to all the extracted coordinates;

searching in an image library according to the obtained function, searching to obtain the name of the article and outputting;

the image library is used for storing the names of the articles and corresponding functions;

when comparing the color value of a pixel with the color values of the neighboring pixels around the pixel, when the differences between the color value of the pixel and the color values of all the neighboring pixels around the pixel are within a set color difference range, respectively obtaining the differences between the color value of the pixel and the color values of all the neighboring pixels around the pixel, calculating the average number of all the differences, accumulating the average number on the color value of the pixel and replacing the original color value of the pixel;

when the difference value between the extracted color value of the pixel point and the color value of at least one adjacent pixel point around the pixel point is not in a set color difference range, changing the color value of the pixel point into a set color value;

when one or more functions are derived from the coordinates,

selecting a function from the image library;

sequentially verifying all the extracted coordinates in the function, and outputting the function when at least two coordinates meet the function;

repeating the above process until all functions in the image library are traversed;

and substituting one numerical value in the coordinates into the corresponding position of the function, and determining that the coordinates meet the function when the difference value between the obtained numerical value and the other numerical value of the coordinates is within a set error range.

2. A computer image recognition method as in claim 1, wherein said image library includes a web crawler module, said web crawler module being adapted to update the contents stored in said image library.

Images