ISS Speed Analysis Dashboard v2

What it does: Filters based on speed percentiles of individual matches with separate top and bottom thresholds
Method: Uses np.percentile() to calculate separate speed thresholds for bottom and top percentiles
Logic: Removes matches with speeds in the bottom X% and/or top Y% of all speeds (configurable separately)
Example: Bottom 5% + Top 10% means it removes the slowest 5% and fastest 10% of all match speeds
Purpose: Removes outlier speeds with flexible control over slow vs fast outlier removal

What it does: Filters based on number of matches per image pair
Method: Counts matches per pair and removes pairs with too few matches
Logic: Removes entire image pairs that have fewer than X matches
Example: 10 means it removes any image pair that has fewer than 10 keypoint matches
Purpose: Removes poor quality image pairs (where feature detection failed)

What it does: Filters based on statistical outliers using standard deviation
Method: Uses np.std() to calculate standard deviation and removes matches beyond Xσ from mean
Logic: Removes matches with speeds more than X standard deviations away from the mean speed
Example: 2.0σ means it removes matches with speeds >2 standard deviations from the mean
Purpose: Removes extreme statistical outliers (very unusual speeds that might be errors)

What it does: Filters based on statistical outliers using Median Absolute Deviation
Method: Uses np.median() to calculate MAD and removes matches beyond X*MAD from median
Logic: Removes matches with speeds more than X*MAD away from the median speed
Example: 3.0*MAD means it removes matches with speeds >3*MAD from the median
Purpose: Robust outlier removal that is less sensitive to extreme values than standard deviation

What it does: Filters based on image brightness and contrast to categorize cloudiness
Method: Analyzes image properties (brightness, contrast) to classify image quality
Logic: Clear images have high brightness/contrast, cloudy images have low brightness/contrast
Example: Clear: brightness ≥120 Filters based on image brightness and contrast to remove cloudy images. Clear: high brightness (≥120) and contrast (≥55). Cloudy: low brightness (≤60) or contrast (≤40). contrast ≥55, Cloudy: brightness ≤60 or contrast ≤40
Purpose: Removes poor quality images (cloudy/overcast conditions that affect ISS visibility)

What it does: Uses machine learning to classify images based on quality
Method: Applies trained ML model to analyze image features and predict quality
Logic: Classifies images as "Good" (high quality) or "Not_Good" (low quality)
Example: Good: clear ISS visibility, Not_Good: poor visibility due to clouds/lighting
Purpose: Removes low quality images using automated ML classification