dimGrid = dim3( (L/dimBlock.x)/8, H/dimBlock.y, 1 ) ;

kernel_convoSepShx8pV<<< dimGrid, dimBlock, 8*dimBlock.x*(dimBlock.y+2*r)*sizeof(char) >>>(d_outc, H, L, r );		  
cudaMemcpyToArray( array_img_inc, 0, 0, d_outc, H*L*sizeof(char) , cudaMemcpyDeviceToDevice) ;
kernel_convoSepShx8pH<<< dimGrid, dimBlock, 8*(dimBlock.x+2*r)*dimBLock.y*sizeof(char) >>>(d_outc, L, H, r );