Image Processing Apparatus Binarization Threshold Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing binarization processing methods for document images struggle to set an appropriate threshold due to the inclusion of out-of-document pixels, which affects the accuracy of binary image generation when the document sheet is not oriented correctly with the scanning area.
Innovation Solution
An image processing apparatus and method that adjusts the histogram by subtracting out-of-document pixels from the luminance values to set a more accurate threshold for binarization, improving the generation of binary image data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If binarization processing is performed using a threshold set based on the histogram of all pixels in the scanning area, then the processing is simple and fast, but the threshold accuracy deteriorates due to the inclusion of out-of-document pixels
Solution Approach 1:
The patent segments the pixels in the scanning area into two distinct groups: document pixels and out-of-document pixels. This segmentation is achieved by detecting the document's position and orientation, then identifying pixels that fall within the document boundaries versus those that fall outside. By separating these pixel groups, the histogram can be generated using only document pixels, thereby improving threshold accuracy without requiring complex additional processing steps beyond the initial document detection
Solution Approach 2:
The patent extracts the out-of-document pixels from the total pixel set by calculating the difference between the scanning area and the detected document area. This extraction process removes the harmful influence of out-of-document pixels (such as those from the scanner background or surrounding environment) from the histogram calculation, allowing the threshold to be determined solely based on actual document content, thus improving measurement precision
2Adaptability or versatility
If the document sheet orientation is not aligned with the scanning area, then scanning flexibility is improved, but the proportion of out-of-document pixels increases, worsening binarization accuracy
Solution Approach 1:
The patent performs preliminary document detection and orientation analysis before conducting binarization processing. By detecting the document's position, orientation, and boundaries in advance, the system can identify and exclude out-of-document pixels from the histogram calculation. This preliminary action ensures that even when the document is scanned at various orientations, the subsequent binarization uses only relevant document pixels, maintaining accuracy while preserving scanning flexibility
Solution Approach 2:
The patent dynamically adjusts the pixel selection parameters based on the detected document orientation and position. Instead of using a fixed scanning area for histogram generation, the system modifies the effective pixel set by applying geometric transformations and masking based on the document's actual orientation. This parameter change allows the system to adapt to different document orientations while maintaining binarization accuracy by excluding out-of-document pixels
Data Source
AI summary
An image processing apparatus includes a controller configured to control an image scanner to scan a document sheet and generate target image data, generate an original histogram of an index value representing a luminance of each pixel forming the target image data, determine whether a particular condition is satisfied that includes a requirement that a difference value between an orientation of the document sheet and a reference orientation in the image scanning process is equal to or larger than a particular value, when the particular condition is satisfied, calculating a count of out-of-document pixels other than pixels representing an image of the document sheet, generate an adjusted histogram by subtracting at least a part of the count of the out-of-document pixels from frequencies of index values within a particular range in the original histogram, setting a specific threshold of the index value for binarization processing, based on the adjusted histogram.