A method, apparatus, electronic device, and storage medium for character recognition are provided. The method may perform image processing on an acquired original image to obtain a region to be recognized. The region may include a character. The method may determine an area ratio of the region to be recognized on the original image. The method may determine an angle between the region to be recognized and a preset direction. The method may determine a character density of the region to be recognized. The method may perform character recognition on the character in the region to be recognized in response to determining that the area ratio is greater than a ratio threshold, the angle is less than an angle threshold, and the character density is less than a density threshold.

“The present disclosure describes a method, an apparatus, and a non-transitory computer-readable medium for detecting sensitive text information such as privacy-related text information from a signal and modifying the signal by removing the detected sensitive text information therefrom. The apparatus receives the signal such as an image, a video clip, or an audio clip, and recognizes a text string therefrom. The apparatus then detects, from the text string, a substring based on a similarity between the substring and a regular expression, and modifies the signal by removing information related to the detected substring from the signal.”

“The present disclosure describes a method, an apparatus, and a non-transitory computer-readable medium for detecting sensitive text information such as privacy-related text information from a signal and modifying the signal by removing the detected sensitive text information therefrom. The apparatus receives the signal such as an image, a video clip, or an audio clip, and recognizes a text string therefrom. The apparatus then detects, from the text string, a substring based on a similarity between the substring and a regular expression, and modifies the signal by removing information related to the detected substring from the signal.”

Methods and systems for extracting personal data from a sensitive document are provided. The system includes a document prediction module, a cropping module, a denoising module, and an optical character recognition (OCR) module. The document prediction module predicts type of document of the sensitive document using a keypoint matching-based approach and the cropping module extracts document shape and extracts one or more fields comprising text or pictures from the sensitive document. The denoising module prepares the one or more fields for optical character recognition, and the OCR module performs optical character recognition on the denoised one or more fields to detect characters in the one or more fields.

An Automated Teller Machine (ATM) for detecting security vulnerabilities by removing noise artifacts from documents receives a transaction request when a document is inserted into the ATM, where the document contains a noise artifact at least partially obstructing a portion of the document. The ATM generates an image of the document, where the image displays at least one data item comprising a sender's name, a receiver's name, and a number representing an amount. The ATM determines whether the noise artifact obstructs at least partially one data item. In response to determining that the noise artifact obstructs at least partially one data item, the ATM generates a test clean image of the document by removing the noise artifact from the image. In response to determining that the noise artifact is removed, the ATM approves the transaction request.

A system for adapting an image noise removal model based on a device processing capability receives, from a computing device, a request to adapt an image noise removal module for the computing device. The system compares a processing capability of the computing device with a threshold processing capability. The system determines whether the processing capability is greater or smaller than the threshold processing capability. In response to determining that the processing capability is greater than the threshold processing capability, the system sends a version of the image noise removal module that is adapted for computing devices with processing capabilities less than the threshold processing capability, where the version of the image noise removal module is adapted to have a number of neural network layers less than a threshold number of neural network layers.

High-accuracy character recognition has not been realized for a document having a space between lines is narrow, a document in which line contact occurs at a plurality of positions, and a document in which a ratio of lines with line contact is high. Noises are removed from divided line images that are obtained by dividing a text image into line units, and the removed noises are added to a neighboring divided text line image, thus restoring the character image which has been divided into the plurality of lines. This realizes the high-accuracy character recognition.

A method for generating learning data for the neural network may comprise generating a license plate image by combining a background image, a frame image and a text image, generating a transformed image by performing at least one of a geometry transformation and a filter transformation on the license plate image, setting a text corresponding to the text image as target data for the transformed image, and generating the learning data including the transformed image and the target data.

The present disclosure relates to systems, non-transitory computer-readable media, and methods for implementing content-aware filters to autonomously remove scan marks from digital documents. In particular implementations, the disclosed systems utilize a set of targeted scan mark models in a scan mark removal pipeline. For example, each scan mark model includes a corresponding content-aware filter configured to identify document regions that match a designated class of scan marks to filter. Examples of scan mark models include staple scan mark models, punch hole scan mark models, and page turn scan mark models. In certain embodiments, the disclosed systems then use the scan mark models to generate mark-specific masks based on document input features. Additionally, in some embodiments, the disclosed systems combine the mark-specific masks into a final segmentation mask and apply the final segmentation mask to the digital document for correcting the identified regions with scan marks.

A system and method. The system may include a display, a lens having distortion, an image generator, and a processor. The lens may be configured to focus light received from an environment. The image generator may be configured to receive the light from the lens and output a stream of images as image data, wherein each of the stream of images is distorted. The processor may be configured to: receive the image data from the image generator; detect a point source object in the stream of images of the image data; enhance the point source object in the stream of images of the image data; undistort the stream of images of the image data having an enhanced point source object; and output a stream of undistorted images as undistorted image data to the display.