Thread 6:EXC_BAD_ACCESS(code 1, dress=0x8) - ios

Question

I am using iPhone5s to do black object tracking: but often meet with
Thread 6:EXC_BAD_ACCESS(code 1, dress=0x8)
and then my App quit suddenly. Could anyone tell me why this happen?
this error happens at :
template<typename _Tp> inline
_Tp Rect_<_Tp>::area() const
{
return width * height; //Thread 6:EXC_BAD_ACCESS(code 1, dress=0x8)
}
//this method is in types.hp in latest opencv framework
my colored object recognition code is as below:
#pragma mark - Protocol CvVideoCameraDelegate
#ifdef __cplusplus
- (void)processImage:(cv::Mat &)image{
Mat imageCopy,imageCopy2;
cvtColor(image, imageCopy, COLOR_BGRA2BGR);
cvtColor(imageCopy, imageCopy2, COLOR_BGR2HSV);
//smooth the image
GaussianBlur(imageCopy2, imageCopy, cv::Size(5,5),0, 0);
cv::inRange(imageCopy, cv::Scalar(0,0,0,0), cv::Scalar(180,255,30,0),
imageCopy2);
/*****************************find the contour of the detected area abd draw it***********************************/
//2-D point to store countour
std::vector< std::vector<cv::Point>> contour1;
//do opening on the binary thresholded image
int erosionSize = 3;
Mat erodeElement =
getStructuringElement(cv::MORPH_ELLIPSE,cv::Size(2*erosionSize+1,2* erosionSize+1), cv::Point(erosionSize,erosionSize));
erode(imageCopy2, imageCopy2, erodeElement);
dilate(imageCopy2, imageCopy2, erodeElement);
//Acual line to find the contour
cv::findContours(imageCopy2, contour1, RETR_EXTERNAL, CHAIN_APPROX_NONE);
//set the color used to draw the conotour
Scalar color1 = Scalar(50,50,50);
//loop the contour to draw the contour
for(int i=0; i< contour1.size(); i++){
drawContours(image, contour1, i, color1);
}
/******END*****/
/****************************find the contour of the detected area abd draw it***********************************/
/****************************Appproximate the contour to polygon && get bounded Rectangle and Circle*************/
std::vector<std::vector<cv::Point>> contours_poly(contour1.size());
std::vector<cv::Rect> boundedRect(contour1.size());
std::vector<cv::Point2f> circleCenter(contour1.size());
std::vector<float> circleRadius(contour1.size());
for (int i=0; i< contour1.size(); i++){
approxPolyDP(Mat(contour1[i]), contours_poly[i], 3, true);
boundedRect[i] = boundingRect(Mat(contours_poly[i]));
minEnclosingCircle((Mat)contours_poly[i], circleCenter[i], circleRadius[i]);
}
/******END*******/
/*****************************draw the rectangle for detected area ***********************************************/
Scalar recColor = Scalar(121,200,60);
Scalar fontColor = Scalar(0,0,225);
//find the largest contour
int largestContourIndex=0;
for (int i=0; i<contour1.size(); i++){
if(boundedRect[i].area()> boundedRect[largestContourIndex].area())
largestContourIndex=i;
}
int j=largestContourIndex;
if(boundedRect[j].area()>40){
rectangle(image, boundedRect[j].tl(), boundedRect[j].br(), recColor);
//show text at tl corner
cv::Point fontPoint = boundedRect[j].tl();
putText(image, "Black", fontPoint, FONT_HERSHEY_COMPLEX, 3, fontColor);
}
// cvtColor(imageCopy, image, COLOR_HLS2BGR);
}
#endif

Answer 1

Finally figure out why :
Just as #SolaWing said, there exist some null pointer. For future Viewers, just want to make it more clear:
Problem is at the following code:
if(boundedRect[j].area()>40){
rectangle(image, boundedRect[j].tl(), boundedRect[j].br(), recColor);
//show text at tl corner
cv::Point fontPoint = boundedRect[j].tl();
putText(image, "Black", fontPoint, FONT_HERSHEY_COMPLEX, 3, fontColor);
}
for this block of code, it already assume that there always exist detected areas, But Actually, when there is no target area in front of Phone Camera, the contour.size() is zero, that being said, for std::vector<cv::Rect> boundedRect(contour1.size()); boundedRect is null pointer, then there will be a problem when I use if(boundedRect[j].area()>40){}, which is using first pointer of the null pointer.