Image Processing Apparatus Alpha Value Smoothing for Depth Cues
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image generation methods produce unnatural images by clearly depicting the border between focused and unfocused objects due to abrupt changes in α values based on depth values, leading to unrealistic rendering of depth cues.
Innovation Solution
An image processing program that adjusts α values for adjacent pixels with significant depth value differences, smoothing and synthesizing images to create a more gradual transition between focused and unfocused areas, thereby blurring the border between objects, resulting in a more realistic depth representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the α value is set simply in accordance with the depth value of each pixel, then the focused image can be generated efficiently, but the border between focused and unfocused pixels appears clearly resulting in unnatural images
Solution Approach 1:
The patent modifies the α value parameter by introducing a correction mechanism. Instead of using only the depth-based α value, the system corrects α values for pixels adjacent to object borders by referencing the maximum and minimum α values in the image, creating a smoothed transition that prevents abrupt boundaries while maintaining focus effects
Solution Approach 2:
The patent introduces an intermediary correction process between depth value calculation and final image synthesis. The correction unit acts as a mediator that adjusts α values based on spatial relationships with adjacent pixels, preventing direct mapping of depth values to α values and thereby avoiding harsh borders
2Measurement precision
If the α value is set to 0 for pixels at the focal point and increased for pixels away from the focal point, then the focused object is clearly depicted, but the border between focused and unfocused objects becomes sharply defined appearing unnatural
Solution Approach 1:
The patent applies parameter transformation by correcting the α value distribution. It calculates the maximum and minimum α values in the image, then uses these to adjust the α values of pixels near borders, creating a gradual transition zone that eliminates sharp boundaries while preserving focus accuracy at the focal point
Solution Approach 2:
The patent introduces dynamic adjustment of α values based on the local image context. Rather than using fixed α values based solely on depth, the system dynamically corrects α values by comparing with adjacent pixels and applying smoothing calculations that adapt to the specific image content and depth distribution
Data Source
AI summary
A computer of an information processing apparatus sets an α value of each pixel in accordance with a depth value (Z value) of the pixel of a reference image. The α value is set such that a synthesizing ratio of the reference image is higher for a pixel having a depth value closer to a predetermined reference value. Next, the computer increases the α value which is set for a pixel having a smaller α value among two adjacent pixels which have an α value difference of a predetermined value or greater. Then, the computer synthesizes the reference image and a blurred image corresponding to the reference image based on the α value which is set for each pixel after being processed by the increasing processing.