Mach-Band effect simulated point light source metal steel seal character detection method

Abstract

The invention relates to a Mach-Band effect simulated point light source metal steel seal character detection method, which comprises the following steps: 1, photographing a to-be-detected metal steelseal by using an area-array camera under a point light source to obtain a frame of input image data; 2, preprocessing the input image data; 3, establishing a self-deduction quotient simulation imagebased on a Mach band; 4, normalizing the value domain of the auto-quotient simulation image by using a mapping function according to an integer or floating point type quantization form; 5, inputting the normalized image into a text detector based on a deep neural network to obtain a text envelope area; 6, performing perspective transformation on the character envelope area, and solving a perspective transformation matrix; 7, inputting the image subjected to perspective transformation into an optical character recognition network to obtain a recognition result of the metal steel seal characters. The point light source is used for detecting the metal steel seal characters, the complexity, the limiting degree and the device scale of the detection device are reduced, the detection accuracy ishigh, the speed is high, and the cost is low.

Description

[0001] (1) Technical field: [0002] The invention relates to a text detection method, in particular to a point light source metal steel stamp text detection method simulated by the Mach band effect. [0003] (two), background technology: [0004] Steel stamps are usually stamped on the outer side of the box of metal parts (such as gearboxes) or on the nameplates on the sides of the motor and internal combustion engine housings. The stamped characters are either raised or depressed. In industrial production or other automation processes, When a metal part is to be used, its identity needs to be identified, that is, it is necessary to detect the text on the steel stamp on the side of the metal part shell. The current detection methods mainly include: [0005] Use a coaxial light source to create the highlight area of ​​the text baseline. Because this method has high requirements for light source debugging, it is usually used on mass production lines, but it has no practical sig...

AI Technical Summary

Problems solved by technology

[0008] For industrial light sources, Gaussian light and coaxial light mean that the equipment is bloated, the detection area is huge, and the footprint is large, which cannot be flexibly embedded in common compact environments

[0009] However, not using an ideal light source can lead to inconsistencies in the high-frequency features of uneven objects under the light source

An extreme example is that a point light source is an extreme form of unbalanced light sources. The current detection system is difficult to overcome the problems of inconsistent highlights, shadows, and high-frequency features produced by point light sources.

At the same time, the existing highly integrated industrial cameras with active light sources cannot be used normally in high-intensity exposure environments

[0010] Therefore, the existing equipment and methods cannot meet the diversity of visual inspection application scenarios

Images