public ProcessResult(IReadOnlyList <Coord> coords, PatternArguments pattern, IReadOnlyList <AnglePrediction> angles, IReadOnlyList <Coord> coordPredictions)
{
Coords = coords;
Pattern = pattern;
Angles = angles;
CoordPredictions = coordPredictions;
}
public static void Main(string[] args)
{
if (args.Length < 3)
{
Console.WriteLine("Argument Error");
return;
}
string patternPath = args[0];
string pointsPath = args[1];
bool useDefault = bool.Parse(args[2]);
PatternArguments pattern = PatternArguments.ReadFromFile(patternPath);
List <Coord>[] eyePoints = ReadPoints(pointsPath);
IProcessor processor;
if (useDefault)
{
processor = new DefaultProcessor();
}
else
{
processor = new TimeTagSensitiveProcessor();
}
DateTime startTime = DateTime.Now;
ProcessResult result = processor.Process(eyePoints[0], pattern);
double cost = GetMatchCost(result);
WriteResult(result);
Console.WriteLine(cost);
Console.WriteLine("======");
result = processor.Process(eyePoints[1], pattern);
cost = GetMatchCost(result);
DateTime endTime = DateTime.Now;
WriteResult(result);
Console.WriteLine(cost);
Console.WriteLine((endTime - startTime).TotalMilliseconds);
Console.WriteLine();
foreach (Coord coord in result.CoordPredictions)
{
Console.WriteLine("{0}, {1}", coord.X, coord.Y);
}
}
public ProcessResult Process(IReadOnlyList <Coord> coords, PatternArguments pattern)
{
int[] lengthStepSum = new int[pattern.LineCount + 1];
double lengthSum = 0;
double[] pointScales = new double[pattern.LineCount + 1];
double[] actualTimes = new double[pattern.LineCount + 1];
{
lengthStepSum[0] = 0;
for (int i = 1; i <= pattern.LineCount; i++)
{
lengthStepSum[i] = lengthStepSum[i - 1] + pattern.LineLengths[i - 1];
}
lengthSum = 1.0 * lengthStepSum[pattern.LineCount];
for (int i = 0; i <= pattern.LineCount; i++)
{
pointScales[i] = lengthStepSum[i] / lengthSum;
}
}
// skip some points in the very beginning and ending
List <Coord> trimmedPoints;
{
int headSkip = HEAD_SKIP;
int endSkip = END_SKIP;
trimmedPoints = coords.Skip(headSkip).Take(coords.Count - headSkip - endSkip).ToList();
}
// take sample points
// get two points that are most close to the corresponding break-point by comparing timeoffset tag
int[,] sampleIndex = new int[pattern.LineCount + 1, 4];
{
double baseOffset = trimmedPoints.First().TimeOffset;
double duration = trimmedPoints.Last().TimeOffset - baseOffset;
for (int i = 0; i <= pattern.LineCount; i++)
{
actualTimes[i] = pointScales[i] * duration + baseOffset;
}
int halfSample = SAMPLE_COUNT / 2;
int count = trimmedPoints.Count - halfSample - halfSample;
int j = halfSample;
bool add;
for (int i = 0; i <= pattern.LineCount; i++)
{
add = false;
for (; j <= count; j++)
{
if (trimmedPoints[j].TimeOffset > actualTimes[i])
{
sampleIndex[i, 0] = System.Math.Max(j - 2, 0);
sampleIndex[i, 1] = System.Math.Max(j - 1, 1);
sampleIndex[i, 2] = System.Math.Min(j, count - 1);
sampleIndex[i, 3] = System.Math.Min(++j, count);
add = true;
break;
}
}
if (!add)
{
sampleIndex[i, 0] = count - 3;
sampleIndex[i, 1] = count - 2;
sampleIndex[i, 2] = count - 1;
sampleIndex[i, 3] = count;
}
}
}
// take {SAMPLE_COUNT} predicted points for each break-point
List <List <Coord> > keyPoints = new List <List <Coord> >();
{
int halfSample = SAMPLE_COUNT / 2;
List <Coord> predicts;
double time;
for (int i = 0; i <= pattern.LineCount; i++)
{
time = actualTimes[i];
Coord a = Math.CalcHelper.PredictCoord(trimmedPoints[sampleIndex[i, 0]], trimmedPoints[sampleIndex[i, 1]], time);
Coord b = Math.CalcHelper.PredictCoord(trimmedPoints[sampleIndex[i, 2]], trimmedPoints[sampleIndex[i, 3]], time);
Coord c = new Coord((a.X + b.X) / 2, (a.Y + b.Y) / 2, time);
predicts = new List <Coord>()
{
a, b, c
};
keyPoints.Add(predicts);
}
}
List <Coord> predictedCoords = new List <Coord>();
foreach (List <Coord> predictCoords in keyPoints)
{
predictedCoords.AddRange(predictCoords);
}
Dictionary <int, int> anglePredicts = new Dictionary <int, int>();
AnglePrediction prediction;
List <AnglePrediction> predictionResults = new List <AnglePrediction>();
List <PredictionRecord> angleRecords;
double ang;
int angleInt;
for (int i = 1; i < keyPoints.Count - 1; i++)
{
anglePredicts.Clear();
foreach (Coord begin in keyPoints[i - 1])
{
foreach (Coord mid in keyPoints[i])
{
foreach (Coord end in keyPoints[i + 1])
{
if (begin.EqualsIgnoringTime(mid) || mid.EqualsIgnoringTime(end))
{
continue;
}
ang = Math.CalcHelper.GetAngle(begin, mid, end);
angleInt = (int)ang;
if (angleInt == int.MaxValue || angleInt == int.MinValue)
{
continue;
}
if (anglePredicts.ContainsKey(angleInt))
{
anglePredicts[angleInt]++;
}
else
{
anglePredicts[angleInt] = 1;
}
}
}
}
angleRecords = new List <PredictionRecord>();
foreach (var pair in anglePredicts)
{
angleRecords.Add(new PredictionRecord(pair.Key, pair.Value));
}
prediction = new AnglePrediction(pattern.Angles[i - 1], angleRecords);
predictionResults.Add(prediction);
}
return(new ProcessResult(coords, pattern, predictionResults, predictedCoords));
}
public ProcessResult Process(IReadOnlyList <Coord> points, PatternArguments pattern)
{
int[] lengthStepSum = new int[pattern.LineCount + 1];
double lengthSum = 0;
double[] pointScales = new double[pattern.LineCount + 1];
{
lengthStepSum[0] = 0;
for (int i = 1; i <= pattern.LineCount; i++)
{
lengthStepSum[i] = lengthStepSum[i - 1] + pattern.LineLengths[i - 1];
}
lengthSum = 1.0 * lengthStepSum[pattern.LineCount];
pointScales = new double[pattern.LineCount + 1];
for (int i = 0; i <= pattern.LineCount; i++)
{
pointScales[i] = lengthStepSum[i] / lengthSum;
}
}
List <Coord> trimmedPoints;
{
int headSkip = HEAD_SKIP;
int endSkip = END_SKIP;
trimmedPoints = points.Skip(headSkip).Take(points.Count - headSkip - endSkip).ToList();
}
int[] sampleIndex = new int[pattern.LineCount + 1];
{
int halfSample = SAMPLE_COUNT / 2;
int count = trimmedPoints.Count - halfSample - halfSample;
for (int i = 0; i <= pattern.LineCount; i++)
{
sampleIndex[i] = (int)System.Math.Floor(pointScales[i] * count + halfSample);
}
}
List <List <Coord> > keyPoints = new List <List <Coord> >();
{
int halfSample = SAMPLE_COUNT / 2;
foreach (int index in sampleIndex)
{
keyPoints.Add(trimmedPoints.Skip(index - halfSample).Take(SAMPLE_COUNT).ToList());
}
}
List <Coord> predictedCoords = new List <Coord>();
foreach (List <Coord> coords in keyPoints)
{
predictedCoords.AddRange(coords);
}
Dictionary <int, int> anglePredicts = new Dictionary <int, int>();
AnglePrediction prediction;
List <AnglePrediction> predictionResults = new List <AnglePrediction>();
List <PredictionRecord> angleRecords;
double ang;
int angleInt;
for (int i = 1; i < keyPoints.Count - 1; i++)
{
anglePredicts.Clear();
foreach (Coord begin in keyPoints[i - 1])
{
foreach (Coord mid in keyPoints[i])
{
foreach (Coord end in keyPoints[i + 1])
{
if (begin.Equals(mid) && mid.Equals(end))
{
continue;
}
ang = Math.CalcHelper.GetAngle(begin, mid, end);
angleInt = (int)ang;
if (angleInt == int.MaxValue || angleInt == int.MinValue)
{
continue;
}
if (anglePredicts.ContainsKey(angleInt))
{
anglePredicts[angleInt]++;
}
else
{
anglePredicts[angleInt] = 1;
}
}
}
}
angleRecords = new List <PredictionRecord>();
foreach (var pair in anglePredicts)
{
angleRecords.Add(new PredictionRecord(pair.Key, pair.Value));
}
prediction = new AnglePrediction(pattern.Angles[i - 1], angleRecords);
predictionResults.Add(prediction);
}
return(new ProcessResult(points, pattern, predictionResults, predictedCoords));
}