Zhu-Takaoka Two-dimensional Pattern Matching

Two dimensional pattern matching.

Details may be added later....

Corresponding more work can be found in Pattern Matching and Text Compression Algorithm, Maxime Crochemore, Thierry Lecroq.

Let's enjoy the code first:

#define REHASH(a, b, h)    (((h - a * d) << 1) + b)

void getNext(char *pattern, int n, int next[]){
	int i = 0, j = -1;
	next[i] = j;
	
	while(i < n){
		while(j >= 0 && pattern[i] != pattern[j])
			j = next[j];
		
		++i; ++j;
		next[i] = j;
	}
}

void bitsCompare(BIG_IMAGE bigImg, SMALL_IMAGE smallImg, int bigRow, int bigCol, int smallRow, int smallCol, int lastRow, int lastCol){
	// The beginning coordinate in big image.
	int i0 = lastRow - smallRow + 1;
	int j0 = lastCol - smallCol + 1;
	
	for(int i = 0; i < smallRow; ++i)
		for(int j = 0; j < smallCol; ++j)
			if(bigImg[i0 + i][j0 + j] != smallImg[i][j])
				return;
	
	// Record the position of successful match.
	OUTPUT(i0, j0);
}

void KMP_Inline(BIG_IMAGE bigImg, SMALL_IMAGE smallImg, int bigRow, int bigCol, int smallRow, int smallCol, int bigImgHashArr[], int smallImgHashArr[], int next[], int lastRow){
	int i = 0, j = 0;
	
	while(i < bigCol){
		while(j >= 0 && bigImgHashArr[i] != smallImgHashArr[j])
			j = next[j];
		
		++i; ++j
		
		// If matched with pattern, then j should be not less than smallCol
		if(j >= smallCol){
			bitsCompare(bigImg, smallImg, bigRow, bigCol, smallRow, smallCol, lastRow, i-1);
			j = next[smallCol];
		}
		
	}
}

void ZT_TwoDimMatch(BIG_IMAGE bigImg, SMALL_IMAGE smallImg, int bigRow, int bigCol, int smallRow, int smallCol){
	int bigImgHashArr[BIG_COL], smallImgHashArr[SMALL_COL], next[SMALL_COL];
	
	// Preprocessing
	// Compute first bigImg hash array
	for(int j = 0; j < bigCol; ++j){
		bigImgHashArr[j] = 0;
		for(int i = 0; i < smallRow; ++ i)
			bigImgHashArr[j] = (bigImgHashArr[j] << 1) + bigImg[i][j];	// The mod we use implicitly here is MAX_INT
	}
	
	// Compute the smallImg hash array
	for(int j = 0; j < smallCol; ++j){
		smallImgHashArr[j] = 0;
		for(int i = 0; i < smallRow; ++i)
			smallImgHashArr[j] = (smallImgHashArr[j] << 1) + smallImg[i][j];	// The mod we use implicitly here is MAX_INT
	}
	
	// Last row of one checking window
	lastRow = smallRow - 1;
	// digit of re-hash
	d = 1;
	for(int j = 1; j < smallRow; ++j)
		d <<= 1;
	
	getNext(smallImgHashArr, smallCol, next);
	
	// Searching
	while(lastRow < bigRow){
		KMP_Inline(bigImg, smallImg, bigRow, bigCol, smallRow, smallCol, bigImgHashArr, smallImgHashArr, next, lastRow);
		
		// Rehash the big hash array
		if(lastRow < bigRow - 1)
			for(int j = 0; j < bigCol; ++j)
				bigImgHashArr[j] = REHASH(bigImg[lastRow - smallRow + 1][j], bigImg[lastRow + 1][j], bigImgHashArr[j]);	// The mod we use implicitly here is MAX_INT
		
		++lastRow;
	}
	
}