Auto-Focusing Method for Point Light Source Scenes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional auto-focus methods under point light source scenes often focus on spurious crests, leading to image blur due to the presence of spurious peaks in image sharpness evaluation curves, resulting in inaccurate and unstable focusing.
Innovation Solution
A focusing method that identifies point light source scenes by analyzing regional luminance variance, partitions the image into regions, and adjusts the focus based on luminance ratios and thresholds to determine the correct focus point, thereby avoiding local extremes and improving focusing accuracy and stability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional auto-focus based on image sharpness evaluation is used under point light source scene, then focusing speed is maintained, but focusing accuracy deteriorates due to spurious peaks in evaluation curve
Solution Approach 1:
The patent performs preliminary detection of point light source scenes before executing the focusing algorithm. By identifying the presence of point light sources in advance through luminance analysis and gradient calculation, the system prepares the appropriate focusing strategy (using luminance variance instead of sharpness evaluation), thereby avoiding the spurious peak problem from the outset and ensuring both accuracy and stability.
Solution Approach 2:
The patent changes the evaluation parameter from image sharpness to regional luminance variance when point light source scenes are detected. This parameter substitution transforms the evaluation curve from having spurious peaks to having a clear unimodal characteristic, enabling reliable focus detection. The system dynamically selects evaluation parameters based on scene characteristics, resolving the contradiction between speed and accuracy.
2Ease of operation
If hill climbing search algorithm is used to find maximum sharpness evaluation value, then focusing process is simple, but it gets trapped in local extremes under point light source scene
Solution Approach 1:
The patent replaces the sharpness evaluation parameter with regional luminance variance parameter when point light source scenes are detected. This parameter change fundamentally alters the evaluation curve shape from multi-peaked to unimodal, allowing the simple hill climbing algorithm to converge to the correct focus point without getting trapped in local extremes, thus maintaining simplicity while improving accuracy.
Solution Approach 2:
The patent converts the harmful effect of point light sources (which create spurious peaks in sharpness evaluation) into a beneficial detection mechanism. By utilizing the unique luminance characteristics of point light sources, the system identifies these scenes and switches to an alternative evaluation method, thereby turning the problematic feature into a useful indicator for selecting the appropriate focusing strategy.
3Productivity
If focus search is performed based on image sharpness evaluation value, then focusing algorithm is straightforward, but repeat dither of focusing effect occurs
Solution Approach 1:
The patent switches the evaluation parameter from image sharpness to regional luminance variance in point light source scenes. This parameter change eliminates the oscillatory behavior and repeat dithering by providing a stable, unimodal evaluation curve that guides the focusing algorithm smoothly to the correct focus point without反复 oscillation, thereby improving both efficiency and stability.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A focusing method is disclosed, comprising: recording image statistical information corresponding to each position of a focusing lens from the beginning of focusing, wherein the image statistical information comprises a luminance value of image; determining whether the current scene is a point light source scene or not based on the recorded image statistical information; and executing a focusing process based on a regional luminance variance of image when the current scene is determined as a point light source scene. Wherein, the regional luminance variance of image denotes a mean of the square of differences between luminance values of a plurality of regions partitioned from an image and an average luminance value of the image. The method of the present disclosure can effectively improve the stability and accuracy of focusing effect under a point light source scene.